Motto : Crapula ingenium offuscat. Traduction : "le bec du perroquet qu'il essuie, quoiqu'il soit net" (Pascal).

Ce blog est ouvert pour faire connaître les activités d'un groupe de recherches, le Séminaire de métaphysique d'Aix en Provence (ou SEMa). Créé fin 2004, ce séminaire est un lieu d'échanges et de propositions. Accueilli par l'IHP (EA 3276) à l'Université d'Aix Marseille (AMU), il est animé par Jean-Maurice Monnoyer, bien que ce blog lui-même ait été mis en place par ses étudiants le 4 mai 2013.

Mots-clefs : Métaphysique analytique, Histoire de la philosophie classique, moderne et contemporaine,

Métaphysique de la cognition et de la perception. Méta-esthétique.

Austrian philosophy. Philosophie du réalisme scientifique.

samedi 4 janvier 2020

Gibson and Gestalt: (Re)Presentation, Processing, and Construction

Gary Hatfield[1]

Abstract. Seeking to avoid the typical binary choices between symbolic representations and no representations, or between functionally decomposable psychological processes and no psychological processes, or between direct perception of mind-independent physical properties and indirect perception of sense data, this article proposes that even a clear-thinking friend of Gibson can accept that perception of the environment is mediated by appearances and that such appearances are produced by functionally decomposable, rule-instantiating psychological processes. In so doing, it avoids both hyper-intellectualization of the perceptual process and the positing of sense data as immediate objects of perception. It considers notions of perceptual mediation from classical Gestalt psychology, while referencing recent phenomenological arguments showing that perception does not simply conform to mind-independent physical properties. Perception presents objects and scenes under a phenomenal aspect, in a manner suitable to be (and evaluable as) action-guiding. Finally, it examines neuroscientific decompositions of Gibsonian information pickup mechanisms, finding that such mechanisms are reasonably described as effecting processes of construction (on a not-necessarily-cognitivist conception of construction). As a separate point, such mechanisms are usefully described as including subpersonal, non-symbolic representations and processes of information integration.

Keywords: James J. Gibson, Gestalt psychology, Ecological psychology, Constructivism, Neural models, Representation

1/  Introduction

Arguments about the role of representation and construction in perception and cognition are often beset by oversimplification into binary choices: representation vs. anti-representation, symbols vs. connectionism, symbols vs. dynamic systems, narrow content vs. ecological embodiment, direct theory vs. constructivism. These dichotomies may be paired up: symbolic representation vs. anti-representational versions of connectionism or dynamicism. Then the argument might go: symbolic representations and syntactically based computational processes are not good for explaining, say, perception, so we should reject computational processes and representations and, according to Gibson (1966b, 1979) and some of his friends (e.g., Chemero 2009), embrace instead direct realism and resonance models of information pickup (resonance being a type of dynamic model).
In reality, the landscape of positions and arguments is not so simple. I want to map a portion of this landscape as I see it. My themes are representation, processing, and construction as related to Gibson, with an eye toward his Gestalt predecessors. I contend that, on some understandings of representation and construction, Gibson’s arguments against them are not compelling. I focus on perception and leave cognition aside for now, which means that I examine the visual perception of the surface layout with some qualities, without going into affordances or object recognition. Gibson himself was deeply interested in how things look, including surface layouts (e.g., 1950, chs. 1, 3; 1966b, p. 205; 1967, pp. 129, 139, 141; 1979, pp. 166, 206–207).[2] I examine that aspect of perception, as available to ordinary consciousness.
Of course, I don’t equate visual perception with conscious visual experience. I only require that such experience often reveals aspects of the environment, including surface layout and qualities, whether directly, as Gibson thought, or via mediating phenomenal experience, as the Gestaltists (especially Koffka) thought and as I think. I offer evidence and theoretical considerations to suggest that visual experience typically does not fully conform to mind-independent physical properties of the layout, but that it nonetheless presents the environment phenomenally in a way that reveals organism-relative stable features that can guide action. 
This sounds like an ecological approach, and so it is intended to be. Still, it holds that the environment is presented under a subjectively conditioned aspect.[3] Further, in recognizing that perceptual experience does not simply copy the environment, this approach fits a generic notion of “construction” in perception: that perception deviates from a neutral physical description of the environment, reflecting the character of perceptual processes. Gibson of course objected to any “constructivist” accounts that posited representations which enter into intelligent constructive processes, finding them to be motivated by an assumption that stimulus information is impoverished, requiring inferential cognitive operations to transform impoverished stimulus values into accurate presentations of the environment. But, I argue, on some accounts that are generically constructivist, neither their motivation nor their essential features require a doctrine of impoverished information or intellectualized cognitive processes. Accordingly, Gibson and clear-thinking Gibsonians could accept that subject-relative phenomenal experience presents (and represents) the environment under a constructed aspect, while avoiding the truly bad aspects of some mediational accounts: sense data[4] and hyper-intellectualization of the perceptual process.
I also reflect on the organismic processes[5] that produce environmentally directed perceptual experience. Gibson distinguished two levels of explanation for perception: a psychological level of perceptual systems that “detect” physical properties of the environment in responding to complex structures of stimulation (Gibsonian information) and a physiological level in which mobile sense organs are looped with patterns of receptor activations and brain activity so that the system physiologically “resonates” to environmental information (1966b, pp. 2–5, 28–29, 158–160, 253; 1979, pp. 52–58). He described the latter in physiological (dynamic) terms, but there may be grounds for introducing a functional division of labor that includes a role for subpersonal represented information about the optic array, or about stimulus patterns, without inherently requiring cognitive notions of inference or interpretation. But, leaving the question of subpersonal representations aside (that is, taking Gibson on his own terms), I maintain that such resonating processes are reasonably dubbed “constructivist” in any case.

2/  Some conceptions of representation and construction in perception

Questions about representation in perception arise in at least two contexts. First: Is perceptual experience of the environment at the personal (or whole-organism) level appropriately described as representing that environment? Let us call this the problem of phenomenal experience of the environment. We can consider two answers already mentioned, Gibson’s direct theory, which eschews representations, and the Gestalt notion of phenomenal mediation, which, with hind-sight, is naturally read as representational at the level of phenomenal experience. Second: Are the processes or detector activities that mediate environmental contact plausibly said to instantiate representational content, that is, represented information-states that interact within and between subsystems, ultimately to yield perceptual experience? Let us call this the problem of information integrationto produce perception. The relevant processes are typically considered to be subpersonal, yielding as their product personal-level phenomenal experience.[6]
Gibson rejected representations in addressing either problem, advancing instead his “direct realism.” He spurned organism-level representations because he thought of them as sense data interposed between perceiver and environment, which become the immediate objects of perception, from which the environment can only be inferred, not seen (1966b, pp. 1–5, 267; 1979, pp. 60, 238). In his view, we perceive the environment without subjectively conditioned phenomenal mediation. I offer instead a conception of visual experience as mediating perception, without rendering that experience into the immediate object of perception and without requiring that the environment be inferred.
As mentioned, Gibson also opposed representation-instantiating subprocesses as needlessly posited to add information beyond that found in optical stimulation, through complex cognitive operations (1966b, pp. 1–2, 267; 1979, p. 238). Accordingly, a second aspect of his saying that perception is direct renders it as cognitively unmediated; no clever cognitive processes, innate or learned, are required (see Hatfield 1988). In his theory of information pickup, perceptual systems respond to stimulus information and thereby detect environmental structures. I propose that, nonetheless, analysis of the detector mechanisms that pick up the information reasonably finds a role for represented information (e.g., about optical structure), which information is not hyper-intellectualized and doesn’t invoke symbol-mediated computation.
Gibson’s view is often contrasted with a “constructivist” opposition (Heft 1981; Norman 2001). The notion that perceptual experience is a construction arises from two main sources, leading to two conceptions of construction. First, the view that perception arises through clever manipulation of impoverished information demands intelligent constructive processes. Second, the fact that perceptual experience sometimes, or perhaps even always, deviates from physical reality is taken for a sign that the experience is constructed. The Necker cube illustrates both points. It is two dimensional but we usually experience it in three dimensions. So, the typical constructivist maintains, the third dimension is added by inferential construction. But, since the image on paper is flat and the experience is three dimensional (with, indeed, two three-dimensional organizations), experience deviates from physical reality and hence includes constructed elements.
The first conception renders perception as cleverly constructed out of impoverished information through a cognitive reconstruction, a kind of best guess about what is present. Accordingly, as I look upon a scene, the basis for my perception is two impoverished images on my retinas (perhaps supplemented by oculomotor sensations). The constructivist psychologist Irvin Rock (1982, 1983) proposed that we see the shapes, sizes, and colors of things by intelligently applying rules to this skimpy information. (We leave aside here the harder problem of classifying and identifying the kinds of things and the individuals in the scene.) Rock allows that this reconstruction might in fact be accurate or veridical (1982, p. 528; 1983, p. 24). However, skeptics charge that if perception is a reconstruction from poor information, it is subject to mistakes and distortions, just as other reconstructive processes are. Memory is one such reconstructive process (e.g., Neisser and Libby 2000). As regards vision, the danger is that constructed perceptions might be mere fabrications or concoctions.
While this first conception has construction make up the difference between the two-dimensional retinal image and three-dimensional experience, the second conception moves from a mismatch between perception and the external physical scene to the conclusion that perception is constructed. Donald Hoffman (1998) promotes this line. He emphasizes mismatches such as those described by the Gestalt principles of organization. These include figure-ground relations, in which figures become more salient than ground even though, physically, they have the same reality as surfaces, as with the black lines and white surface of the Necker-cube line drawing. They also include cases in which the same physical shape yields different perceptual organizations, as when the cube shifts orientation (a “Gestalt switch”). The perceived change in the orientation of the cube is a subjective addition with no corresponding physical reality, producing two different three-dimensional misperceptions of the same two-dimensional black lines. Hoffman concludes that, because perception is constructed in these respects, it might be a mere construction more generally, that is, a fabrication or concoction (1998, ch. 8).
Let us consider more closely the ideas of Rock, Hoffman, Gibson, and, as needed, the Gestalt psychologists, especially Kurt Koffka (1935), on representation and construction. As noted, the first two authors agree that perception is a construction but disagree on what this entails for veridicality. Gibson (1966b, 1979) argues that because visual information, when properly analyzed, is able to specify the physical properties of the scene, there is no need for construction; usually, perception veridically presents the physical properties of a mind-independent reality.[7] At the same time, he acknowledges that deviation from accurate perception of the physical layout, as in cases of incomplete size or shape constancy (e.g., elliptical pennies or converging railway tracks) “seem to pose a very real difficulty” for his theory (1966b, p. 306; also, 1979, p. 166). Intermediate cases of constancy, he suggests, indicate that perception doesn’t track environmental reality. As he understands things, constructivists attribute such results to an incomplete transformation (construction) of visual sensation into perception (a conception he seeks to avoid, as discussed below).
Hoffman and Gibson agree that any deviation of phenomenal experience from the mind-independent physical environment is evidence of construction (even if Gibson holds that such deviation is aberrant). We can call this shared conception of construction PDE:

PDE (Phenomenal deviation from the physical environment): The deviation of phenomenal experience from the mind-independent physical environment indicates perceptual construction.

As we shall see, the Gestalt psychologists, whom Hoffman and Gibson each invoke, agree with PDE in principle, without using the terminology of “construction.” Hoffman and Gibson both hold that such deviation is always misperception. Here I disagree, as would the Gestaltists.
Because I disagree, I want to draw a different map of the relations among these positions and the implications of accepting that perception is constructed. To do this, I must broaden the notion of construction. I want to include all cases in which the visual system must transform its input in order to produce perception. I thus endorse CT:

CT (Construction as transformation): Any aspect of perception that differs from proximal stimulation (or from the structure of the ambient optic array at a point of observation, still or moving) is to be deemed a transformation and hence a construction.

CT broadens the notion of construction to include, as a transformation and hence a construction, any case in which what is perceived regularly differs from the structure of the energy distributions received at the sense organs or sampled by perceptual systems. Rock agrees already. The Gestaltists (Koffka 1935, pp. 129–141; Köhler 1947, pp. 132–133; see Epstein 1994, p. 186) happily recognized that perceptual processes in vision respond to the two-dimensional stimulus mosaic by producing an experience in three dimensions, which presents the world under a specific aspect (including organizational features such as figure-ground). Gibson claimed to avoid perceptual construction, except in deviant cases, because the optic array specifies the spatial layout (1966b, ch. 14).
CT may seem so broad as to be weak. My job is to show that, under this broad notion, surprising instances of construction reveal themselves. I will show that such transformations and constructions must occur even for Gibson, so that, under CT, his theory is constructivist (independent of the question of subpersonal representations). Indeed, it is only from the implausible perspective of naïve realism that perception just copies, or directly encompasses, physical things (on which, see Hatfield 2016). Any description of perceptual processes that pays attention to the characteristics of optical stimulation will note that what the eyes and the visual system receive or respond to, no matter how rich it may be in information, does not copy the world and so must be transformed to yield perception (a point brought home by descending into the physiology of information pickup).
Further, I maintain that perceptual deviations from external physical properties need not be characterized as misperceptions, or even as non-veridical. Here, the standards of veridicality track presentations of the environment in a way that is adequate for the guidance of action:

AA (Aspect for action): Perception presents the environment under an aspect suitable for action guidance; deviation from mind-independent physical properties need not be a “mistake” if perception effectively supports action.

I offer phenomenal grounds for believing that ordinary spatial perception does not match a mind-independent description of the physical scene, but presents that scene under a subjectively conditioned aspect. A similar point was made by the Gestalt psychologists.
So viewed, perception is more deeply constructed than is usually thought. Perception includes a phenomenal construction that differs from a bare representation of mind-independent reality, which need not be a bad thing. Indeed, for guiding action, a subject-dependent, environment-tracking phenomenal construction may be what is needed.
In subsequent sections, this paper argues that spatial perception is constructed in this way and should be so regarded even by Gibsonians. It offers a notion of radical construction that is neutral about the mechanisms by which the construction occurs. Construction can be pervasive, even if perceptual processes are not inferential.

3/  Organism-level experience of an environment: Traditional constructivism, Gibson, and Gestalt

Gibson considered the accepted theory of visual perception, which I am calling traditional constructivism, to consist of these key points: 

(1C) The stimulus for vision, including that for the spatial configuration of a scene, is impoverished and hence ambiguous, failing to specify the third dimension; 
(2C) this impoverished stimulus produces “sensations” that mirror the proximal stimulus (the retinal image in the case of vision); 
(3C) in order to make up for this impoverishment and produce perceptions that correspond (or come close) to the physical scene, computations or inferences, perhaps based on memory or past experience, are required; 
(4C) the product of sensations and inferences is a constructed sense datum, a mental item that serves as the object of perception; 
(5C) the proximal sensations and inferential processes are unconscious or unnoticed (which is why we don’t discover them easily by phenomenal reflection).[8]
Accordingly, the perceptual system infers or cognitively constructs the environment from impoverished visual stimulation, usually described as a static retinal image.

To replace this picture, Gibson emphasized that in normal acts of perception, the perceiver and the perceiver’s eyes are free to move, yielding dynamic changes in the optic array that offer rich stimulus information. He held that:

(1G) The stimulus for vision, the ambient optic array for a stationary or mobile observer, carries information that unequivocally specifies the spatial layout; 
(2G) this rich stimulation supports direct perception of the spatial layout with its objective properties; 
(3G) no computations or inferences are needed; instead, the rich stimulus information is simply picked up by perceptual systems; 
(4G) we do not, in normal veridical perception, perceive sense data or have mediating experiences but are directly aware of the physical scene itself; 
(5G) there is no need for unconscious psychological operations; rather, there are unconscious physiological processes that integrate across receptors and organs.
This construal of Gibson’s theory fits both his 1966 and 1979 books (even if the later book more fully develops some ideas).[9]

An additional contrast between traditional constructivism and Gibson concerned experimental methods. Gibson noted that evidence for proximally oriented sensations typically arose from experiments using impoverished stimulus conditions (e.g., immobile observers, monocular viewing, brief presentations). Under such conditions, size and shape constancy break down so as to reveal, according to constructivism, the true elements of perception. According to Gibson (1966b, ch. 14; 1972, pp. 225–226), impoverished stimulus conditions produce abnormal sensations. But under normal, rich stimulus conditions, including a mobile perceiver using both eyes, the information received by the perceptual system yields direct awareness of the spatial layout and other properties of things (1966b, pp. 3–4, 287–304; 1979, pp. 70­–71, 149, 160–161).
Gibson often wrote as if these were the only two choices. If you speak of things looking small in the distance, or of a penny looking elliptical when seen at a slant, you are forcing proximal sensations into experience, either through impoverished viewing conditions or by adopting a special attitude. Gibson acknowledged that if perception did not regularly conform to the “objective facts” (1966b, pp. 6, 306) of the environment, that would be a problem for his theory (1966b, p. 306; 1979, pp. 166–167). Problematic examples (for him) included cases in which, under full-cue (full-information) conditions (including a mobile observer), there is “incomplete perceptual constancy” (1966b, p. 306; 1979, p. 160), as when a coin is said to look elliptical or railway tracks to converge phenomenally. In these full-cue cases, he suggested that the described result either reflects poor phenomenology or results from an attitude which creates sensations that intrude upon experience. A standard example is a pictorial attitude, which can induce an experience of a two-dimensional visual field (1950, pp. 16, 33–38; 1966b, pp. 235–238; 1979, pp. 3, 71). 
Gibson countered the poor phenomenology by saying that the penny looks like a circle at a slant so that, unless viewed at an extreme angle, it looks like a circle (1966b, p. 306). I find this response compelling. He responded to the claim that things look small in the distance by saying that they don’t. He cited an experiment with stakes in an open landscape in which average responses of full constancy were achieved at a distance of nearly one-half mile (1950, pp. 183–186; 1979, p. 160). If this means that, under ordinary perceptual conditions, there is no phenomenal diminution at distances of a quarter mile – or of a hundred feet, or twenty feet – I am incredulous. I don’t believe it.[10] I am not saying that, to an adult observer, things look as if they were diminished in physical size in the distance. Nor am I denying that one can alter the experience of size by adopting a pictorial attitude. But I am asserting that, in normal perception, without the pictorial attitude, things are diminished phenomenally in the distance.
Let us return to the phenomenological point anon. Gibson also recognized a theoretical approach distinct from traditional constructivism: that of the Gestalt psychologists. Koffka was Gibson’s colleague at Smith College, whose weekly seminar he attended from 1928 to 1941. Koffka’s Principles of Gestalt Psychology, praised by Gibson as “one of the great books of the century” (1967a, p. 131), laid out a theory of perception that accepted (1C), the impoverishment of the stimulus with respect to perceptual experience, but rejected (2C) to (5C). Koffka’s position may be summarized as follows:

(1K) The stimulus for vision, including that for the spatial configuration of a scene, is impoverished and hence ambiguous, failing to specify the third dimension; 
(2K) this impoverished stimulus sets constraints on central dynamic processes; 
(3K) in order to make up for this impoverishment and produce perceptions that correspond (or come close) to the physical scene, holistic field processes operate in the brain in relation to stimulus constraints; 
(4K) the product is three-dimensional organized experience that tends toward constancy, and which mediates between perceiver and the physical environment by phenomenally presenting that environment under an aspect of organization; 
(5K) we learn of the characteristics of the physiological field processes from the characteristics of experience (Koffka 1935, chs. 3–6; see also Köhler 1947, chs. 3–5).[11]

The Gestaltists were no friends of proximal sensations or unconscious mental operations.[12] The impoverished stimulus is able to yield a presentation of a spatial and chromatic world through brain processes that operate dynamically to produce organized perception, tending in the direction of the constancies of shape, size, and color. Organization includes figure-ground relations, enhanced contrast at edges, grouping by proximity or similarity, and more. Perceptions in the direction of shape constancy result from field forces in the physiological medium that, for example, drive a slim ellipse toward more regularity (roundness), under the constraint of the shape-slant invariance hypothesis (1935, pp. 229–233). Size is somewhat different. Koffka accepts Wolfgang Köhler’s suggestion that distance is undervalued; objects are “brought nearer” and made smaller (for dynamic reasons), in accordance with the size-distance invariance relation (Koffka 1935, pp. 229, 237). (Size-distance invariance is explained below.)
The Gestalt psychologists distinguished physical objects from the perceptual experience of those objects. The physical objects are not in direct contact with consciousness as in the “acquaintance” relation of naïve direct realism,[13] but our perceptual experience “mediates” contact with the physical environment. Perceptual experience is not composed of sensations or constructed through inferences but is produced by dynamic brain processes that operate holistically to yield an organized percept. Koffka called the world of immediate experience the behavioral environment, as opposed to the geographic or physical environment. In his example, a horseman on Lake Constance rides over solid ground in his behavioral environment but over the thin ice of the lake in his geographical environment. Koffka noted that his “behavioural environment” has similarity with Köhler’s “direct experience,” the latter not meaning direct realism but phenomenologically described experience as we have it. Koffka preferred his own term over Kohler’s because “it signifies the exact place which it has in the system, viz., the mediation between geographical environment and behaviour” (1935, p. 36).[14]
Figure 1 represents the behavioral environment (BE) in relation to the geographical environment (G), the real organism (RO), and the ego and its phenomenal behavior (PH B). Real behavior (RB) is directed toward and regulated by BE, and it affects G, thereby changing BE (G “produces” BE). BE, RB, and PH B are, “in some sense,” in RO (1935, p. 40). Although one might take BE to be a kind of sense datum, I don’t read Koffka that way. Rather, I take BE to be a presentation to the ego of G under an aspect.[15] It presents the world in a subjectively  

Fig. 1 Koffka’s representation of the relations among the behavioral environment (BE), the geographical environment (G), real behavior (RB), phenomenal behavior (PH B), and the real organism (RO). Koffka (1935, p. 40), public domain

conditioned way. Further, the Gestaltists did not believe that in order to be correct the perceptual representation (BE) had to present only mind-independent physical structures. Rather, BE presents G under aspects of figure-ground organization, enhanced edges, groupings, and the like. They did not consider these departures from a neutral physical description to be illusions or mistakes. The behavioral environment of the rider is, however, mistaken as to the character of the surface he traverses (it is ice, not solid ground). I’m less clear on whether Koffka considered departures from constancy, such as things looking smaller in the distance or pennies looking elliptical, to be mistakes. My sense is that he would agree that standards of correctness for size perception need not majorize physical size in itself; for instance, the comparative phenomenal “nearness” of distant things might be functionally good because it supports the perceptual surveyability, clarity, and articulation of those things.[16]
I believe that Gibson did think of these departures from physics as mistakes. His ecological psychology provided for “ecological” facts, such as: the scale of the environment in relation to the organism; the placement of the organism in relation to a scene; standing features of illumination and of surfaces; and the relation of an animal to earth as a supportive substratum and to media such as water or air. Such factors are part of his important doctrine of organism–environment mutuality. But, I contend, in applying this doctrine, Gibson retained a latent physicalism, which took metric physical values for size and shape to be the target of perception.
Organism–environment mutuality describes organism and environment in relation to one another in various ways, including objects in relation to surfaces and organism-relative functionally significant structures, such as shelters and paths (Gibson 1979, chs. 1–3). Nonetheless, it also privileges physical measurements and descriptions. While offering a variety of classifications of ecologically relevant units, it accepts that, at the relevant scale, the organism responds to and perceives physically objective sizes, shapes, and other properties (spatial layouts, media, substances). Gibson thus retained his earlier (1966b) notion that the “objective” is constituted by ordinary physical properties, so that the sensory systems he wants to investigate are those “by which an organism can take account of its environment and cope with objective facts” (p. 6). The standard for normal spatial vision is “the perception of the true layout” (p. 306). In his definitional discussion of organism­–environment mutuality (1979, pp. 8–9), one primary difference concerns the scale of the animal. While physics (as a discipline) focuses on the scales of atoms and galaxies, ecological psychology uses millimeters and meters to describe animal-relevant environmental properties (which physics, as a discipline, does not describe). By retaining a physically accurate “match” as the standard of successful size perception (pp. 160–161), Gibson did not escape, as he might have, the naïve realist and physicalist preference for an “objective” physical world that perception simply reveals (albeit at an organismic scale).

4/  Organism level experience of an environment: Things looking small in the distance

Let us return to size perception and the phenomenology of things looking small in the distance. Recall that Gibson believed that this phenomenology is artificially produced by taking a pictorial attitude so as to experience a two-dimensional perspective image of the scene. He illustrated what is involved through Figure 2. By manipulating our attitude toward the rectangle on the ground, we can produce an intermediate size and shape. But, he claims, with full-information conditions and no attitude, we should perceive the figure where it is with its true size and shape. He held that, in the normal case, there is information to specify the true sizes of  

Fig. 2 Equivalent configurations from a stationary, monocular point of observation. The rectangle on the ground plane creates a solid visual angle. The same structure is consistent with intermediate trapezoids of various slants in relation to the line of sight, including a perspective projection into an upright plane. Gibson (1979, p. 167), used with permission

things. For example, an object on the ground covers the same amount of (homogenous) ground texture independently of distance. This invariant relation between ground texture and size provides the information for size constancy (Gibson 1979, p. 163). Or, for a person standing and looking at a row of telephone poles of equal height on flat terrain, poles near and far are cut by the horizon line at the same proportional location; given that eye-height is known, information for the physical heights of the poles is present through proportional relations (p. 165).
Gibson claimed that invariant ratios within the optic array specify the “objective,” that is, physical, layout (1979, pp. 72–73, 126, 129, ch. 9). For a mobile observer using both eyes, the information “specifies” the spatial configuration of the environment (sizes, shapes, distances, etc.), and the visual system simply needs to “pick up” the relevant information,[17](allegedly) yielding direct, unmediated awareness of the scene (1966b, pp. 223, 266–267; 1967b; 1972, pp. 215–216; 1979, pp. 72–73, chs. 9, 14).
There are things to challenge conceptually in Gibson’s claim. For instance, one might wonder how the presence of information sufficient to specify the layout should produce a direct unmediated acquaintance with physical objects, of the sort envisioned (problematically, I think) by naïve direct realism. But here I prefer to mount a phenomenological challenge. I believe that Gibson has used binary thinking in assimilating the phenomenology of intermediate constancy to a case of intruding sensations. On the contrary, phenomenological reflection, also supported by experimental evidence, suggests that intermediate results are quite normal.
Let’s consider size constancy and things seeming smaller in the distance. Koffka (1935, p. 229) held that the size-distance invariance hypothesis (not his term) covers this case. The SDIH doesn’t itself specify what size something will appear to have, but it says that, holding visual angle constant, the perceived size of an object directly varies with its perceived distance. If an object is perceived as being at its true physical distance, it should be perceived with its true size; if it is perceived as nearer than it is, it will be phenomenally smaller (Figs. 3a and 3b). Koffka offered this relation as a phenomenal description; he was not invoking a quasi-cognitive underlying process in which perceived distance combines with perceived visual angle.
The perception of railway tracks extending into the distance, or of a long, straight hallway in a building, offer counter-instances to Gibson’s phenomenology. I perceive railway

Fig. 3 Perceived size in relation to perceived distance. (a) Poles of equal height at different distances; visual angle (retinal projection) diminishes with distance. If perceived distance matches physical distance, then, by the SDIH, perceived size matches physical size. (b) SDIH applied to two objects, A and C, that are at different distances but have the same visual angle. Perceived size varies with perceived distance; if distance is under-perceived, A might appear at B and C at E. Author

tracks as converging (Fig. 4, left). This convergence is not merely, as Gibson supposes, that found in linear perspective. If it were, I would experience the tracks as in a two-dimensional perspective projection, which would mean – for the usual perspective plane perpendicular to the line of sight – that they would appear perpendicular to the ground and would project into an upright, rather truncated triangle (Fig. 4, right). But my experience looking down the tracks is of 

Fig. 4 Railway tracks perceived as converging into the distance. The photograph (left) gives a sense of depth; on real tracks, the sense of depth and distance is even more salient. If our experience matched a perspective projection, it would be of an upright triangle (right). Left, public domain; right, author

them receding into the distance (in the third dimension). They converge, but they do so as they extend toward the horizon. And, indeed, there is good evidence that, in normal perception, perceived distance falls short of physical distance in a regular manner, yielding phenomenally converging tracks (and things looking small in the distance). This evidence includes psychophysical studies in the laboratory (Hillebrand 1902; Blumenfeld 1913) and observations taken in landscape settings (Gilinsky 1951; Wagner 1985; Erkelens 2015), with both eyes open and mobile head and eyes.[18]Further, recent developmental studies support a distinction between (a) perceived space, in which objects appear phenomenally smaller in the distance, and (b) learned cognitive responses to such appearances, which sustain judgments that target metric or physical size (Granrud 2012).
In order to understand the regularity of the phenomenal contraction of visual space, consider Figure 5, from Aage Slomann (1968). The drawing shows a rectangular space, say, a 

Fig. 5 An illustration of the effects on phenomenal visual space if distance is under-perceived. For observer P1 (ad), the ground (dc) appears to rise (de), and objects appear closer and smaller (bc appears at gf). From Slomann (1968, Fig. 1), used with permission

hallway, with a physically flat floor Ff running from observer P1’s feet (at d) to staff bc. Slomann reports that we do not experience the floor as “flat” in the sense of forming a ninety-degree angle with the observer’s body (or being perpendicular to gravity), but that the floor phenomenally rises. I agree.[19] And if the floor rises while the directions to locations on the physical floor remain the same (that is, while we continue to perceive directions veridically, or nearly so, as I think we do), then distance must be under-perceived. For example, location c appears in the direction given by the line ac, but because it is seen in relation to the floor, it appears in location f, at the shorter distance af. Taking this phenomenally rising ground to occur regularly, then, in accordance with the SDIH, objects will appear closer and smaller, as physical staff bc appears at phenomenal location gf. The space is contracted in the in-depth dimension, along lines of sight. There is, in fact, no geometric-optical reason why the ground couldn’t appear flat and things appear where they are at their true sizes, or with full phenomenal size constancy (e.g., staff bc at location bc), but that would be quite strange for us: it would mean that the train tracks don’t appear to converge, or that a building a mile away isn’t phenomenally of less extent than the same building 30 feet away.
This diminution, although related to linear perspective, is not the same as a projection into a frontal plane, since phenomenal structures retain three-dimensional depth. It may be regarded as a phenomenal presentation of the physical or geographical environment under an aspect in which distance contracts in a regular manner. Under such a contraction, much useful information is phenomenally available: phenomenal direction is congruent with physical direction; objects retain an invariant proportion to their surrounds across different distances and phenomenal sizes, so that Gibson’s horizon invariant and ground-texture occlusion are preserved as proportions; next-to-ness relations on surfaces are preserved; so are ordinal relations along the ground plane; and so on (see also Warren 2012).
These findings challenge Gibson’s phenomenology. Size constancy is intermediate (see also Daoust 2017). Shape, at the scale of a hallway, is also affected (floors in a long rectangular hallway appear as elongated trapezoids), and train tracks appear to converge.[20] As Gibson realized, this favors phenomenal mediation (1966b, p. 306). If the object and its properties are not directly present in experience (by acquaintance), but are presented in a manner that is subjectively conditioned, then the natural grounding for presented object-properties is subjectively conditioned phenomenal experience. The Gestaltists and Gibson agreed on this (even if Gibson sought to limit applicability by setting full phenomenal constancy as the norm). They accept an age-old logic for drawing a distinction between appearance and reality: that properties as present in experience deviate from mind-independent physical properties. This need not, by the way, commit one to appearances as sense data. Appearances can be conceived as presenting the physical world under a subjectively conditioned aspect (Hatfield 2016). The world is not physically located in the appearances or inferred from the appearances but is presented through the appearances. The world is the object of perception, the appearances the mode of perceiving that object (see also Chisholm 1957, ch. 10).
Contrary to Gibson’s version of direct realism, we don’t perceive the world just-as-it-is physically, or even just-as-it-is relative to our own physical position and physical size.[21] This raises the following question about size: Even if Gibson is right that in ecologically normal situations information sufficient to specify physical size and distance is present, why should the physical value be the target of perception? What is wrong with contracted space, if it is ecologically adequate, and perhaps preferable, for guiding action? Size perception can be about relations, directions, and proportions, presented perceptually in a way that guides action. It needn’t be about mind-independent sizes and distances, although these can be ascertained from the layout as perceived, through known operations of measurement or estimation.[22]
This conception regards perception as having the function of representing, or presenting, relevant features of the environment, including surface layout, in useful ways that allow for normative evaluation. If, for whatever reason, a perceptual experience distorts the invariant object/surround relation, or fails to present visual direction accurately (veridically), these would be misperceptions, of the sort that Gibson viewed as perceptual deficiencies.
The following four features apply to visual experience, conceived as representing, or presenting, the environment under an aspect:

(A) A task analysis (functional decomposition) of vision as a system for environmental contact assigns to visual experience the function of representing the spatial structure of the environment (among other functions), whether metrically or proportionally; 
(B) Such experience is normatively evaluable as veridical or non-veridical;
(C) Experienced visual space presents the environment under an aspect that is subject-dependent; a prime instance is the phenomenal contraction of visual space with distance, with sizes (and surrounds) varying proportionately with distance;
(D) Such presentations are not sense data; they are not the objects of perception; they are the means through which the environments that they present are seen or perceived (in the veridical case).
Gibson would alter (A) to specify direct perception of the objective metric structure of the physical environment. He would accept (B) but reject (C) and (D), suggesting that the system uses rich stimulus information to perceive the physical environment as it is metrically. He is caught out here by the phenomenal fact that things do appear smaller in the distance.

Gibson was a pioneer in conceptualizing environmentally embedded perceptual systems (1966b, ch. 3) and bringing evolution to bear (ch. 9). But in making veridical perception of mind-independent physical properties the norm (for the spatial layout), his task analysis mistakenly retained physicalist leanings regarding the aim of perception and its conditions of adequacy. In contrast, I suggest that a subject-relative contraction of visual space can sustain action-guidance. A regular diminution in the apparent sizes of things can still mediate effective action. Proportional sizes and shapes, and even physical properties such as visual direction, can be assimilated into a subjectively conditioned visual space, in a way that allows for intersubjectively shared, environment-relative standards of veridicality (see Hatfield 2003a, 2016).
If we accept the phenomenal intermediacy of size perception, then we should accept that perception is a construction, in accordance with PDE:

PDE: The deviation of phenomenal experience from the mind-independent physical environment indicates perceptual construction.

The Gestaltists might accept that phenomenal organization, while beneficial, can deviate from mind-independent physical reality and so fits construction broadly construed. Rock and Hoffman accept PDE. Gibson denied phenomenal deviation in normal cases, which is too bad for him.

5/  Perception as radically constructed: Transformation of stimulus information

Let us now consider perception as constructed according to CT, construction as transformation:

        CT: Any aspect of perception that differs from proximal stimulation (or from the structure of the ambient optic array at a point of observation, still or moving) is to be deemed a transformation and hence a construction.

There are obvious instances, as in Rock’s conception that ambiguous two-dimensional stimulus information must be transformed by intelligent perceptual processes to yield the experience of surfaces in three dimensions. Hoffman allows that construction in perception transforms retinal patterns into organized percepts. The Gestaltists accepted something similar, while disagreeing with Rock and Hoffman that the processes must be inferential or cognitively clever. Gibson would allow that such transformation occurs with illusions. Otherwise, he would deny that construction occurs in everyday full-cue perception of a three-dimensional world. My first task is to show that, leaving aside the phenomenal criterion for construction (PDE), that is, putting aside for now my phenomenal argument, Gibson and his followers could accept this broader (hence more radical) notion of construction in perception without abandoning ecological psychology.
Let us continue granting Gibsonian claims about rich information. There is still a need for some mechanism or process to transform stimulus information into perception of the visual world. The information available to the visual system does not copy the world, rather it specifies it in relation to a system that is attuned to, and resonates with, the available information.
An example will help. Gibsonians have analyzed optic flow patterns that produce perception of observer-motion through the environment (Warren 2008; Uesaki & Ashida 2015). When an observer moves forward in a static environment with normal optical texture, that motion produces a flow pattern of outward radial expansion at each eye (Fig. 6). Picking up this information results in an experience of forward motion. There is, however, an obvious difference 

Fig. 6 An optic flow pattern produced by the forward motion of an observer through a stationary setting with normal optical texture. Optical features expand from the point of fixation in the scene. From Palmer (1999, p. 227), used with permission

between expansion within the optic array, an expansion in two dimensions that is sampled by the two-dimensional surface of the retina, and the phenomenal experience of moving forward. Gibson recognized this, and offered a division of labor to handle it. As a psychologist, he worked at the level of perceptual systems and the analysis of stimulus structures. If he found a psychophysical relation between optical expansion and the perception of forward motion, he then said that the perceptual system detects the motion by picking up the information that specifies it.[23] If pressed on how this occurs, he allowed that receptors respond to light energy, but maintained that it was not his job to determine how the receptors and visual nervous system integrate patterns of stimulation. He was ready to move on after finding complex stimulus information that specifies the physical environment (1950, p. viii; 1966b, pp. 2–5, 167; 1979, pp. 53, 251, 263).
Nonetheless, Gibson allowed that neuroscientists might ask how receptor systems respond to and integrate the information in optic flow (1979, p. 251). Neuroscientists answered by seeking neural mechanisms for detecting such flow. An early effort was van de Grind (1988). He proposed motion-detecting neural mechanisms organized in a ring (see also Duffy 2004). For global optic flow, the detectors might be set in concentric rings centered on the fovea. Motion detectors are velocity and direction specific. These detectors would be oriented so as to respond to local motion of optical features moving outward on radial lines, while increasing in velocity and traversing the concentric rings perpendicularly. Van de Grind suggested a similar mechanism for more localized optical expansions providing the information that an object, such as a soccer ball, is approaching the perceiver’s head (see also Frost & Sun 2003). These two mechanisms take as input local or global optical expansion and produce the perceptual experience of a looming object or of the perceiver moving forward.[24] By CT, these are instances of construction as transformation.
A Gibsonian might object that there is no construction because there need be no underlying “intelligent” or hyper-intellectualized processes. And indeed, I haven’t characterized any processes as symbolic, conceptual, inferential, or the like, the usual suite of words associated with Rockian construction. Rather, I have rendered the criterion for construction neutral between Rock and Gibson. The point is to define construction as occurring when perception results from a transformation in relation to received light structures or patterns. We have a transformation in the case of optic flow patterns, which are not themselves three dimensional but yield the perception of motions in three dimensions (observer motion or ball motion). This sort of construction is needed even with Gibson’s rich information. In accordance with neutrality, any “computations” involved might be effected by nonconceptual connectionist nets or by Rockian intelligent inferences (see Hatfield 1990). CT doesn’t specify. Accordingly, cognitive constructivism, as with Rock or Hoffman, becomes but one subspecies of construction as transformation.

6/  Non-cognitive processes of transformation

Van de Grind (1988) approached the neurophysiology of pickup mechanisms using the notion of a “smart mechanism” from Runeson (1977). A smart mechanism accomplishes a task by its mechanical structure rather than by having internal symbols and explicitly represented formulae. Runeson’s example was the planimeter, a device for measuring the area of a closed plane figure. If one arm of the device is anchored and another is used to trace the boundary of the figure completely (one revolution), then the device provides a read-out of the figure’s area. At no point does it consult or explicitly represent a formula for computing area. There is no internal symbol system, hence, the reasoning goes, no representations.
Van de Grind sought to get inside special purpose smart mechanisms, such as those that guide a fish’s swimming or that respond to optic flow in the perception of self-motion or looming. He developed the neural side of smart mechanisms as a contribution to Gibsonian direct theory, as contrasted with “inferential” or “mediated” accounts based on the computer analogy.
Van de Grind conceded that computers can be programmed to act like smart mechanisms: to undertake special purpose information detection. But computer implementations differ from biological implementations of smart mechanisms in that, at bottom, their operations are carried out in a symbol system defined in relation to a CPU. He then compared a direct account of the guided action of swimming to a “mediated” or symbolically implemented model:

If one studies sensorily-guided bodily actions, as most adherents of the “direct perception” theory do, it is sensible to attempt to circumvent postulates about internal representations, mediation, etc. The information relevant to the sensorily-guided bodily actions resides in the environment, the body form, the interface between body and environment and the smart mechanisms tuned to the peculiarities of the situation. No independent symbolic (conventional) representation is needed to replace the “real things,” because the latter are always there and ready to go when needed. (1988, p. 167)

 Here we have a common move. Representational accounts are exemplified by symbolic GOFAI models, in which the perceiver carries an internal model of the environment as opposed to being tuned to the environment itself (and, if we accept Johansson’s point, note 16 above: as opposed to operating on the background engineering assumption that ecological regularities hold within the environment).[25] On this conception, representations take the place of real things in perception. But, the argument goes, there is no need to consult explicit representations of body and environment and to compute from them:
It is more sensible in this case to use the metaphor “tuning” to the environment than to speak of “computations” based on sense-data. The swimming programs are smart special-purpose systems that need no internal model of the body, the environment and the hydrodynamic processes. They are an integral part of the whole system as it has evolved in millions of years and the parts need no “representation” of the other parts of the system, only the appropriate feedback/feedforward signals. The intelligence is distributed, not stored in neurons.[26] (p. 167)
Nonetheless, van de Grind recognized a temptation to treat organismic processes that implement smart mechanisms as carrying out computations or implementing algorithms that account for the “distributed” intelligence of guided swimming. But he disparaged such talk as “far fetched”: 

It seems rather far-fetched to call such natural processes “mediated” or “computational” or to call the swimming programs in the brain “algorithms.” If one chooses to do so, it is hard to see why an old-fashioned balance or a wind-mill should not be called an “algorithm” or why the bending of a tree in the wind should not be called a computation. In my opinion natural law-governed processes should not be called computational or algorithmic. (p. 167)

Of course, the operations of a CPU are also “natural law-governed processes”; van de Grind would, I think, argue that because standard digital computers are designed, and the treating of their outputs as representations rests on convention, they can be separated from natural organic processes (p. 168). He here appeals to the distinction between rule-following and rule-described processes that was common at this time, where rule-following processes consult explicitly symbolically encoded rules, and rule-described processes are so characterized merely because their behavior accords with the laws of nature.
Van de Grind sought to alleviate the temptation to treat natural systems as embodying representations and computations by offering three examples as reductios: the balance, the windmill, and bending tree branches. On the teleofunctional conception of representation that guides my inquiry here,[27] we can rule out branches bending in the wind because we won’t ascribe a function to compute or to represent to this natural system. The bending follows natural laws that describe it, but the bending is not part of a system in which it functions to yield information. By contrast, the windmill is an artifact designed to carry out an intended function. But its teleofunction, van de Grind might agree, is not to compute, but to provide power for grinding wheat or cutting logs. Finally, the balance may be interpreted as a teleofunctional system that computes weights, including adding weights together as they are placed on one pan to even out the to-be-weighed item on the other pan. Analog computers are also computers, even if they don’t have internal symbols (Hatfield 1988). Such systems are rule described (as are all the systems in question). They are not rule following, that is, explicitly-rule-consulting processes. Still, when described functionally they are normatively evaluable: a balance can be broken. We can call such teleofunctionally described processes rule-instantiating (Hatfield 1991).
Indeed, van de Grind (1988, p. 173) continued to feel the pull of describing natural systems as “computing,” which he did in a subsequent example – but, since it was not symbolic computing, he added scare quotes. The example concerned modular smart mechanisms, which he found to be functionally decomposable because intramodular structures explain how the module works; and, also, to be functionally coordinated so as to cooperate (p. 168).
His main example of such mechanisms was the system for detecting optic flow patterns already mentioned (1988, pp. 171–178). Such a system includes, as subcomponents, local (retinal) motion detectors or what van de Grind called bilocal velocity detectors. These have further subcomponents, including orientation sensitive edge-detectors, coincidence detecting cells, and functionally relevant differential response latencies. In his model, these velocity detectors “code” velocity information on the “labelled line” principle: activation of the line of a detector tuned to one velocity sends on to subsequent detector mechanisms the information or content that that velocity has occurred. Various mechanisms might make use of these subcomponents. Thus, a ring of such detectors might detect outward optic flow by being sensitive to simultaneous activation of many such detectors with orientation sensitivities aligned as perpendicular to radii from a common center of expansion. If the ring is centered in the fovea and responds to the whole field, it is a self-motion detector; smaller concentric rings not necessarily centered on the fovea would make looming detectors.[28]
Van de Grind of course considered these detectors to be non-computational and nonrepresentational because they aren’t instantiated in a symbol system. Like Gibson, he offered a binary choice: symbolic representation or no representation. Finding a CPU-driven symbol system to be neurally implausible, he went for no representations and computations. Still, he used teleofunctional language in describing the systems, and presumably he would allow that they can be broken or dysfunctional. And he used the representation-friendly language of “coding” for his velocity detectors (1988, pp. 168, 173). Indeed, it seems natural to regard the velocity detectors as providing, to the higher mechanism, a representation of local (retinal) velocity (or, in middle and late Gibsonian terms, of local changes in the optic array as sampled by the optic-flow system). These higher mechanisms might function to represent a looming object or perceiver motion. But the subunits that feed these higher units are not themselves detecting the environmental layout. Rather, they are tuned to features of proximal stimulation that then enter into further processes to yield a perception of environmental and bodily happenings.[29]
It is reasonable to apply the language of representation and computation in these cases because there is information or content about stimulus patterns that is combined according to rules that are neurally implemented to produce perceptions of the environment. This is a case of rule instantiation, since the rules are sanctioned by a task analysis and are neurally instantiated but without a symbolic medium.[30] Instead of Gibson’s psychological level in which information in the optic array (say, local optical expansion) simply yields detection of changes in the layout (a thing is looming), we have (a) optical information as structure in the light; (b) encoding of aspects of this structure from proximal patterns; (c) integration of representations of this pattern to yield a representation of the looming thing. At the neural level, we cannot simply treat the optical information as “picked up,” as if there is no structure within the pickup mechanism. Rather, the early motion detectors respond, not to the Gibsonian invariant as specifying looming, but to optical patterns including local edges and timed coincidences, yielding the encoding of bi-local velocity in an organized pattern. This description is functionally that of neurally instantiated psychological processes that represent proximal patterns so that they can be integrated by further processes that yield organism-level perceptions. Given what must go into producing such perceptions, and given the difference between optic flow and the perception of forward motion, one might well regard the latter as a construction based on the former. 
The teleofunctional aspect of rule-instantiation also supports ascriptions of incorrect encoding of velocity, leading to improper looming perceptions and or inaccurate perceptions of perceiver motion. The question of whether the information in the optic array “fully specifies” looming becomes moot. What we need to know is whether the information is sufficient for the various detectors to respond accurately in cases of looming or forward motion. The ambiguity or lack of ambiguity of stimulus information must be judged in relation to the mechanisms that respond to the optic array and produce a perception. 
We can summarize the difference between a functionally rule-instantiating perceptual system such as I have described and van de Grind’s conception of a traditional symbol-based system. In the symbol-based system, we have these components:

(1) A description of the perceptual function to be explained, e.g., detection of a fast approaching thing or presentation of the perceiver’s forward motion;
(2) Analysis of the rules for combining represented information (or represented stimulus patterns) from receptor-inputs so as to achieve the result;
(3) A syntactically characterized internal symbol system replete with an internal “reader” that responds to symbols in accordance with their syntax;
(4) Expression in this symbol system of an algorithm for combining the information; 
(5) A neural implementation of the symbol system (reader and symbols). 

 By contrast, a rule-instantiating account of information flow can get by with:

(1) and (2) remain the same;
(3) to (5) are rejected and replaced with:
(3ʹ) Specification of the neural elements available to be combined so as to carry out the receptor-detection of local informational structure or stimulus patterning;
(4ʹ) Psychological description of the neural mechanisms that integrate the represented information to yield the perceptual outcome.

With symbol-systems, (4) and (5) involve both psychological and neural descriptions; so do (3ʹ) and (4ʹ). Earlier, van de Grind (1984, p. 435) glossed “encoding” with “representation,” without embracing symbols; rather, he used the term “representation” to indicate the type of local content that neural mechanisms produce, accepting that this content describes the stimulus aspects to which they are attuned when regarded from a teleofunctional and psychological perspective.
Van de Grind was tempted to think of pickup mechanisms as involving subpersonal representations. I have drawn out these aspects of his position and related them to rule-instantiating processes. Does this vitiate the insights of ecological psychology? It enriches that psychology. The point about rich stimulation is not excluded (but of course remains subject to examination). The mechanisms for information pickup can now be dissected. Organism–environment mutuality can be further articulated. One species of seer may be tuned to different features of the stimulus array than another. For example, flies may not be landscape surface-perceivers but may nonetheless respond to environmental structures as “landable” (Marr 1982, pp. 32–34). Organism-relative aspects of the environment remain a chief object of perception. The arguments in these two sections do not depend on PDE and do not require mediating appearances. Still, the conclusions of these sections – that perception is, in a generic sense, constructed, and that there are neural and psychological decompositions and functional characterizations of the process of perception that are non-cognitive but still invoke representations[31] – are compatible with mediating appearances. The subpersonal representational processes that pick up optic flow may be conceived as yielding the experience of moving forward in a (phenomenally) spatially contracted environment that preserves visual direction.

7/  Embodiment

I’ve sought to overcome binary dichotomies often on offer, for example, between symbolic representations and no representations. More specifically, I’ve argued that an ecologically oriented, Gibson-appreciating theorist can accept notions of organism-level representation as subjectively conditioned phenomenal presentation and can also accept subsystem-encoded representation of local informational structure as part of a functional analysis of how rich stimulation is responded to and processed to yield perception. 
I haven’t directly mentioned embodiment, which opens up a further wide field of discussion (see Shapiro 2014). If embodiment means that the capacities of the whole organism in its ecological setting are considered when theorizing, then the view promoted here is embodied. It assumes that sensory systems are tuned to environmental regularities. Unlike anti-representational views, it allows that such regularities enter into perceptual processing by being instantiated as rules, perhaps in the form of background engineering assumptions, but also as learned regularities. Such ascriptions of representations and their interaction is based in a task analysis and a functional decomposition of the mechanisms that carry out the task. 
Symbolic representations are not needed for this. Sense data are not required. There can be an ecological psychology that decomposes smart mechanisms using notions of coding and representation. This psychology can be described as constructivist, on the grounds that the mechanism must transform optical structure (such as expansive flow) to yield personal-level perceptions (of self-motion). These mechanisms can be analyzed as combining information via internal representations, but the processes involved need not be intellectualized inferences. There can be construction with representations and with mediating appearances but without sense data or inferences. Gibson can rest easy.

Blumenfeld, W. (1913). Untersuchungen über die scheinbare Grösse im Sehraume. Zeitschrift für Psychologie und Physiologie der Sinnesorgane65, 241–404.
Brewer, B. (2011). Perception and its objects. Oxford: Oxford University Press.
Chemero, A. (2009). Radical embodied cognitive science. Cambridge, Mass.: MIT Press.
Chirimuuta, M. (2015). Outside color: Perceptual science and the puzzle of color in philosophy. Cambridge, Mass.: MIT Press.
Chisholm, R. (1957). Perceiving: A philosophical study. Ithaca: Cornell University Press.
Clementz, F. (2000). La notion d’aspect perceptif. In: P. Livet (Ed.), De la perception à l’action (pp. 17–57). Paris: Vrin.
Cormack, L. K., Czuba, T. B., Knöll, J. & Huk, A. C. (2017). Binocular mechanisms of 3D motion processing. Annual Review of Vision Science3, 297–318.
Daoust, L. (2017). Seeing things as we do: Ecological psychology and the normativity of visual perception. Ph.D. Dissertation. University of Pennsylvania.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, Mass.: MIT Press.
Duffy, C. J. (2004). The cortical analysis of optic flow. In: L. M. Chalupa & J. S. Werner (Eds.), The visual neurosciences (Vol. 2, pp. 1260–1283). Cambridge, Mass.: MIT Press.
Erkelens, C. J. (2015). The extent of visual space inferred from perspective angles. i-Perception6, 5–14.
Epstein, W. (1993). The representational framework in perceptual theory. Perception & Psychophysics53(6), 704–709.
Epstein, W. (1994). “Why do things look as they do?”: What Koffka might have said to Gibson, Marr, and Rock. In S. Poggi (Ed.) Gestalt psychology: Its origins, foundations and influence (pp. 175–189). Florence: Olschki.
Frost, B. J., & Sun, H. (2003). The biological bases of time-to-collision computation. In H. Hecht & G. J. P. Savelsbergh (Eds.), Time-to-contact (pp. 13–38). Amsterdam: Elsevier - North-Holland.
Gibson, J. J. (1947). Motion picture testing and research. Washington, D.C.: U.S. Government Printing Office.
Gibson, J. J. (1950). The perception of the visual world. Boston: Houghton Mifflin.
Gibson, J. J. (1959). Perception as a function of stimulation. In S. Koch (Ed.), Psychology: A study of a science (Vol. 1, pp. 456–501). New York: McGraw-Hill.
Gibson, J. J. (1966a). The problem of temporal order in stimulation and perception. The Journal of Psychology62(2), 141–149.
Gibson, J. J. (1966b). The senses considered as perceptual systems. Boston: Houghton Mifflin.
Gibson, J. J. (1967a). James J. Gibson. In E. G. Boring & G. Lindzey (Eds.), A history of psychology in autobiography (Vol. 5, pp. 125–143). East Norwalk, Conn.: Appleton-Century-Crofts.
Gibson, J. J. (1967b). New reasons for realism. Synthese17(1), 162–172.
Gibson, J. J. (1972). A theory of direct visual perception. In J. R. Royce & W. W. Rozeboom (Eds.), The psychology of knowing (pp. 215–240). New York: Gordon and Breach.
Gibson, J. J. (1979. The ecological approach to visual perception. Boston: Houghton Mifflin.
Gilinsky, A. S. (1951). Perceived size and distance in visual space. Psychological Review58, 460–482.
Granrud, C. E. (2012). Judging the size of a distant object: Strategy use by children and adults. In G. Hatfield & S. Allred (Eds.), Visual experience: Sensation, cognition, and constancy (pp. 13–34). Oxford: Oxford University Press.
Hatfield, G. (1988). Representation and content in some (actual) theories of perception. Studies in History and Philosophy of Science19, 175–214. 
Hatfield, G. (1990). Gibsonian representations and connectionist symbol-processing: Prospects for unification. Psychological Research52, 243–252.
Hatfield, G. (1991). Representation and rule-instantiation in connectionist systems. In T. Horgan and J. Tienson (Eds.), Connectionism and the philosophy of mind (pp. 90–112). Boston: Kluwer.
Hatfield, G. (2003a). Objectivity and subjectivity revisited: Color as a psychobiological property. In R. Mausfeld & D. Heyer (Eds.), Colour Perception: Mind and the Physical World (pp. 187–202). Oxford: Oxford University Press.
Hatfield, G. (2003b). Representation and constraints: The inverse problem and the structure of visual space. Acta Psychologica114, 355–378.
Hatfield, G. (2007). The reality of qualia. Erkenntnis66, 133–168.
Hatfield, G. (2012). Phenomenal and cognitive factors in spatial perception. In G. Hatfield & S. Allred (Eds.), Visual experience: Sensation, cognition, and constancy(pp. 35–62). Oxford: Oxford University Press.
Hatfield, G. (2013). Perception and sense data. In M. Beaney (Ed.), Oxford Handbook of the History of Analytical Philosophy (pp. 948–974). Oxford: Oxford University Press.
Hatfield, G. (2016). Perceiving as having subjectively conditioned appearances. Philosophical Topics44(2), 149–178.
Heft, H. (1981). An examination of constructivist and Gibsonian approaches to environmental psychology. Population and Environment, 4(4), 227–245.
Hillebrand, F. (1902). Theorie der scheinbaren Grösse beim binokularen Sehen. Denkschrift der Kaiserlichen Akademie der Wissenschaften Wien, Mathematisch-Naturwissenschaftliche Classe72, 255–307.
Hoffman, D. D. (1998). Visual intelligence: How we create what we see. New York: Norton.
Johansson, G. (1970). On theories for visual space perception: A letter to Gibson. Scandinavian Journal of Psychology11, 67–74.
Koffka, K. (1935). Principles of Gestalt psychology. New York: Harcourt, Brace.
Köhler, W. (1947). Gestalt psychology: An introduction to new concepts in modern psychology. New York: Liveright.
Kosslyn, S., & Hatfield, G. (1984). Representation without symbol systems. Social Research51, 1019–1045.
Lombardo, T. J. (1987). The reciprocity of perceiver and environment: The evolution of James J. Gibson’s ecological psychology. Hillsdale, N.J.: Lawrence Erlbaum.
Marr, D. (1982). Vision. San Francisco: Freeman.
Michaels, C. F., & Palatinus, Z. (2014). A ten commandments for ecological psychology. In L. Shapiro (Ed.). The Routledge handbook of embodied cognition(pp. 19–28)New York: Routledge.
Neisser, U., & Libby, L. K. (2000). Remembering life experiences. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 315–332). Oxford: Oxford University Press.
Noë, A. (2004). Action in perception. Cambridge, Mass.: MIT Press.
Noë, A. (2005). Real presence. Philosophical Topics33, 235–264.
Norman, J. (2001). Two visual systems and two theories of perception: An attempt to reconcile the constructivist and ecological approaches. Behavioral and Brain Sciences25(1): 73–96.
Ooi, T. L. & He, Z. J. (2007). A distance judgment function based on space perception mechanisms: Revisiting Gilinsky’s (1951) equation. Psychological Review114, 441–454.
Palmer, S. E. (1999). Vision Science: Photons to Phenomenology. Cambridge, Mass.: MIT Press.
Reed, E., & Jones, R. (1982). Introduction to Pt. 4, Implications of ecological realism. In E. Reed & R. Jones (Eds.). Reasons for realism: Selected essays of James J. Gibson (pp. 295–302). Hillsdale, N.J.: Lawrence Erlbaum.
Rock, I. (1982). Inference in perception. In P. D. Asquith & T. Nickles (Eds.), PSA: Proceedings of the biennial meeting of the Philosophy of Science Association, volume two: Symposia and invited papers (pp. 525–540). East Lansing, Mich.: Philosophy of Science Association.
Rock, I. (1983). The logic of perception. Cambridge: Mass.: MIT Press.
Rogers, S. (2000). The emerging concept of information. Ecological Psychology12, 335–343.
Runeson, S. (1977). On the possibility of smart perceptual mechanisms. Scandinavian Journal of Psychology18, 172–179.
Searle, J. (1992). The rediscovery of the mind. Cambridge, Mass.: MIT Press.
Shapiro, L. (2011). Embodied cognition. London: Routledge.
Shapiro, L. (Ed.). (2014). The Routledge handbook of embodied cognition. New York: Routledge.
Slomann, A. (1968). Perception of size: Some remarks on size as a primary quality and “size constancy.” Inquiry, 11, 101–113.
Uesaki, M., & Ashida, H. (2015). Optic-flow selective cortical sensory regions associated with self-reported states of vection. Frontiers in Psychology Accessed 28 June 2018.
van de Grind, W. A. (1984). Decomposition and neuroreduction of visual perception. In A. J. van Doorn, W. A. van de Grind, & J. J. Koenderink (Eds.), Limits in perception (pp. 431–494). Utrecht: VNU Science Press.
van de Grind, W. A. (1988). The possible structure and role of neuronal smart mechanisms in vision. Cognitive Systems, 2(2), 163–180.
van Leeuwen, C. (2015). Hierarchical stages or emergence in perceptual integration? In J. Wagemans (Ed.), The Oxford handbook of perceptual organization (pp. 969–988). Oxford: Oxford University Press.
Wagner, M. (1985). The metric of visual space. Perception and Psychophysics38, 483–495.
Wagner, M. (2006). Geometries of visual space. Mahwah, N.J.: Erlbaum.
Wagner, M., Hatfield, G., Cassese, K., & Makwinski, A. N. (2018). Differentiating between affine and perspective-based models for the geometry of visual space based on judgments of the interior angles of squares. Vision2, 1–22.
Warren, W. H. (2008). Optic flow. In T. Albright & R. H. Masland (Eds.), The senses: A comprehensive reference, Volume 2, Vision II (pp. 219–230). Amsterdam: Elsevier.
Warren, W. H. (2012). Does this computational theory solve the right problem? Marr, Gibson, and the goal of vision. Perception41(9), 1053–1060.
Wittgenstein, L. (1969). Philosophische Untersuchungen=Philosophical investigations, 3rd ed. New York: Macmillan.

[1] Department of Philosophy, University of Pennsylvania, Philadelphia, PA 19104-6304  USA  ORCID: 0000-0003-0876-7073

Published in Synthese; special issue on: Gestalt Phenomenology and Embodied Cognitive Science. The final publication is available at

Acknowledgements: An earlier version was presented at The World in Us: Gestalt Structure, Phenomenology, and Embodied Cognition, University of Edinburgh, July 2017; I thank the audience for their queries. Thanks also to referees and issue editors for advice. Larry Shapiro made helpful suggestions subsequently. An abridged version was presented in Aix-en-Provence, December 2018, with stimulating comments by Jean-Maurice Monnoyer and François Clementz. Portions were presented at a workshop on The Philosophy and Psychology of Visual Space, Ohio State University, February 2019; thanks to participants for helpful discussion, and especially to Jim Todd, Bill Warren, and Harry Heft (who sometimes agreed with me, sometimes not).

[2] Gibson’s interpreters highlight the theory of affordances as developed after his (1966b). Reed and Jones (1982, p. 300) write: “Over the last 15 years of his career, Gibson united his various contributions to psychology into a single coherent framework or ecological approach. An idea fundamental to this approach is the concept of affordance.” But Gibson did not abandon inquiry into how things look. His 1979 book opens with: “This is a book about how we see.” He asks how we see, respectively, “the environment around us”; “its surfaces, their layout, and their colors and textures”; “where we are in the environment”; “whether we are moving or not, and, if we are, where we are going”; “what things are good for” (i.e., affordances); and “how to do things” (includes affordances). These questions end with Koffka’s query, “Why do things look as they do?” (1979, p. 1; Koffka 1935, p. 76). Gibson’s attitude toward spatial aspects of perception developed; instead of the “spatial world” of 1950 (p. 8), we have the “layout” with its ecological properties in 1979 – properties that include surfaces and changes in their layout, color, texture, and existence (p. 94). Affordances are important but not all determining. In summarizing his new position, Gibson (1979, ch. 8, p. 143) began: “The notion of information in ambient light to specify affordances is the culmination of ecological optics.” But it is not the only chapter of that optics; optical information relates to surfaces and objects (ch. 5), events (ch. 6), and self-perception (ch. 7). Affordances fit within a broader ecological psychology: “The notion of invariants that are related at one extreme to the motives and needs of an observer and at the other extreme to the substances and surfaces of a world provides a new approach to psychology” (1979, p. 143).
[3] The notion of subjective here is subject-dependent. Some may prefer “subject-relative” or “organism-relative” as the adjective. Such “subjective” aspects of experience can sustain intersubjective agreement and so can be “objective” in that sense; the meaning of “subjectivity” here is not “idiosyncratic variation” (see Hatfield 2003a).
[4] Sense data were introduced by G. E. Moore and Bertrand Russell in the early twentieth century as immediately known objects of perception, distinct from mind-independent physical objects as normally understood. They initially regarded such data as mind-independent “third things” (in addition to perceivers and ordinary objects), but others view them as mind-dependent perceptual contents. (On this history, see Hatfield 2013.) Critics of traditional constructivism take it to posit two-dimensional sense data as the immediate object of visual perception (e.g., Lombardo 1987, pp. 319–320), including Gibson himself (1950, pp. 13, 183, 187; 1966b, pp. 48, 287; 1979, pp. 54, 60–61, 280). Sense data as the immediate objects of perception are now out of favor (with me, as well).
[5] To avoid misunderstanding about Gibson’s opposition to certain kinds of “processing” in visual perception, I observe that while Gibson (1966b, p. 2; 1979, pp. 149–150, 221) was opposed to positing cognitive processes as underlying perception, he regularly described information pickup as a process of the visual system (1966b, pp. 251, 267; 1979, pp. 2, 147, 150, 263); his discussions of such processes often alluded to the physiology of sensory systems (1966a; 1966b, pp. 3, 5, 253; 1979, pp. 56–57, ch. 14, esp. p. 251), even if he did not delve deeply into physiology.
[6] It is convenient to use phenomenal descriptions of human experience. Gibson did this, too. We each recognize that animals perceive and that evidence concerning their perceptions, and also human perception, can be obtained by studying the information available to perceptual systems and its use to guide behavior in natural and experimental settings (see Gibson 1966b, pp. 49–50, ch. 12; 1979, pp. 7, 206–208).
[7] A philosophical counterpart to Gibson’s direct realism is so-called naïve direct realism, which says that no construction is needed because the world with its physical properties is directly present to consciousness via unmediated acquaintance. Having detailed the shortcomings of such theories elsewhere (Hatfield 2016), herein I merely allude to their problems.
[8] This description of traditional constructivism accords with Gibson’s characterization (1966b, pp. 2–5; 1979, pp. 2–4, 251–253); it fits definitions of “constructivism” by Heft (1981, pp. 228–229) and Norman (2001, pp. 74–75). Heft adds that constructivism regards stimulation as lacking value and meaning, which are also constructed, whereas Gibson includes value and meaning in the optical information available to perceivers, a point we leave aside, as we are not focusing on affordances. Finally, the list of standard constructivist positions includes the theories of George Berkeley, Hermann von Helmholtz, Rock, and modern information-processing approaches that invoke inference.
[9] Gibson (1950, ch. 7) assumes an active observer and mentions the “flow of neural excitations” that yields temporally continuous visual experience (117). The 1950 book differs from Gibson (1966b) and (1979) in making retinal images (albeit dynamic images) fundamental, as opposed to the optic array defined at a stationary or mobile point of observation. The latter books make mobile observers the norm: “we must perceive in order to move, but we must move in order to perceive” (1979, p. 223). As Rogers (2000) has observed, static arrays can specify an environment, a point that Gibson reflected on (e.g., 1966b, pp. 198–199; 1979, pp. 250–251). In this article, the phenomenal examples brought against Gibson typically assume a mobile observer.
[10] I am not rejecting Gibson’s empirical results. However, they arose from “objective” instructions, to “estimate the real height” of a stake “out there in the field” (Gibson 1947, p. 203); Gibson (1950, p. 175) described this as a “naïve attitude.” He would have understood “apparent” instructions (to describe the phenomenal or apparent size of the stakes) as directing subjects to report the two-dimensional visual field (1950, ch. 1) – what he subsequently called a “perspective appearance of the terrain and the objects” (1979, p. 161) as contrasted with full constancy responses from “naive observers” (p. 160). His method asked for objective judgments, not responses to the stakes as they appeared.
[11] Point 4K does not fully apply to Köhler, who treated “direct experience” and its objects as the Gestalt (organized, holistic) equivalent of sense data, that is, as the immediate object of perception (1947, pp. 7, 22, 24, 28).
[12] On the relations among Gibson, the Gestalt psychologists, and Rockian constructivism, including a sense in which Gestalt accounts are constructivist, see Epstein (1994).
[13] Good expositions of “acquaintance” in naïve direct realism can be found in Brewer (2011) and Noë (2004, pp. 72–73), even if Noë avoids the term “acquaintance” (2005, p. 258).
[14] More generally, Koffka acknowledged that his “behavioural environment” did not fully capture Köhler’s “direct experience” or even his own notion of a psychological or environmental field (1935, pp. 36–53). He nonetheless found the distinction between geographical and behavioral environments useful, and I use the contrast as he drew it. Still, this distinction was embedded in a longer discussion that qualified and limited it, ultimately giving the geographical environment a significant psychological role (pp. 53–74). Koffka also argued that the distinction does not presuppose a mind–body dualism that places the behavioral environment “in the mind” (pp. 47–48, 66–67).
[15] The behavioral environment is primarily caused by the geographical environment (Koffka 1935, p. 74). The relation between them allows for “questions of true cognition and adequate or adapted behaviour” (p. 68). I render this by saying that the behavioral environment presents the geographical environment under a subjectively conditioned aspect, with more or less accuracy. Subjectively conditioned aspects include presentation of the environment from a point of view, or as possessing subject-dependent characteristics such as phenomenal color and, as discussed below, a contracted visual space. (On color as a subject-dependent aspect, see Chirimuuta 2015, ch. 1; Hatfield 2003a, 2007.) “Seeing under an aspect” was not a technical term for Koffka, but he frequently used “aspect” with the relevant connotation (e.g., 1935, pp. 243, 265). Wittgenstein (1969, pp. 193–208) famously discussed seeing different aspects, as with the duck–rabbit, mostly involving seeing and recognizing: seeing it as a duck, as a rabbit – although he allowed cases in which previous experience of the differing aspects is not needed (p. 207). More recently, Dretske (1995, pp. 30–32) used the term “aspectual shapes” to suggest that we always see objects as being a certain way: e.g., red, round, and stationary; he adopts this language from Searle (1992, pp. 131–133, 155–158). Clementz (2000, p. 22) responds to a problem that arises if Dretske is saying that all seeing is “seeing as” in a conceptual sense. Clementz distinguishes three forms of seeing as and contends (correctly, I think) that Dretske could accept all three: a purely perceptual or sensory-representational seeing as, which is non-conceptual, e.g., representing things visually as having a determinate shape and color; and two forms of “categorizing” seeing-as, one that categorizes based on similarity but is not strictly conceptual (if being conceptual implies entering into judgments), and a conceptual form, in which to see a thing as an X, one must possess the concept X (pp. 36, 38). He also describes appearing under an aspect that is not “seeing as” but amounts to things being perceived under a sensory “mode of presentation” (pp. 38–39). We need not investigate the difference between this notion and non-conceptual seeing as. Herein, unless otherwise stated, appearing under an aspect accords with non-conceptual instances of seeing as or of mode of presentation.
[16] This is partly right; on the functional advantages of a contracted visual space (without full phenomenal constancy, which has things closer and smaller than they are physically), see Hatfield (2016), pp. 170–171.
[17] For now, I am granting that simple “pick up” is enough. But I am sympathetic to Johansson’s contention that additional assumptions are needed if the optic array is to “specify” the distal layout, including that of a fixed environment that does not globally deform; as Johansson put it, “Without a priori assuming the existence of 3-D rigidity there is no specific information about space available in the visual stimulus, even in connection with locomotion. The visual apparatus has to introduce a set of decoding principles for data treatment with the assumption that the expanding and changing proximal patterns represent motion in a rigid 3-D world” (1970, p. 71). (Johansson also advanced a motion-friendly perceptual psychology.)
[18] The literature on the structure of visual space is large. A main topic concerns the form of the visual contraction, whether it is a type of perspective projection (see the next paragraph), which might be Euclidean or non-Euclidean, or an affine compression; see Wagner (2006), chs. 6–7, Ooi & He (2007), Hatfield (2003b, 2012), and Wagner et. al. (2018) for overviews and discussion. Some authors hold that the contraction of distance (so things look small in the distance) is owing to poor distance information (e.g., Rock 1982, p. 533); others find that the contraction may be ecologically beneficial (e.g., Hatfield 2016, pp. 170–171). The contraction might be thought of as a phenomenal accommodation to decreasing information with distance; if the visual system aimed for full phenomenal constancy, then with increasing distance it would need to present like phenomenal units using ever-decreasing information as contained in unit areas of retinal stimulation or solid angles in the optic array. Interestingly, even a committed Gibsonian such as Bill Warren describes this contraction as a “distortion” (2012, p. 1056), suggesting that Euclidean structure is the standard of undistorted space (a position he in fact seeks to avoid). I call these presentations of physically Euclidean structures in a contracted manner a “transformation,” not a distortion.
[19] For psychophysical confirmation, see Ooi & He (2007).
[20] Gibson frequently discussed the train tracks. He assimilated their apparent convergence to the linear perspective structure of the “visual field” (1950, pp. 35–38), while asserting that size constancy characterizes the visual world out to nearly one-half mile (pp. 183–186). Later, he wrote: “I have argued that the coin does not always look elliptical and that the tracks do not always seem to converge (surely not to the locomotive engineer!). But I do not seem to win this argument, for most people say they do, and the results of experiments on shape constancy and size constancy commonly bear them out” (1966b, p. 306). Within his binary framework of physical objectivity and sensation-based introspection, he averred that “visual sensations obtrude themselves on the perception of the true layout and cause the illusion of seeing partially in perspective. Putting it another way, sometimes we attend to the pictorial projections in the visual field instead of exclusively to the ratios and other invariants in the optic array” (p. 306). Accordingly, he must treat phenomenal convergence as resulting from the pictorial attitude. In my view, perception proper presents the physical layout under an aspect of contraction, without need for a special attitude. Adopting an attitude might alter our experience of a planar perspective projection, but no attitude is needed for the phenomenally pervasive appearance of converging hallways and railway tracks (see also Hatfield 2003b, 2012). Finally, observer-motion does not matter in this case. The studies finding a contracted visual space typically employ observers with mobile head and eyes. In my own experience, the railway tracks still exhibit phenomenal convergence even if I am walking. Anyone can test this by walking along a long, straight walkway and noting any affect on phenomenal convergence.
[21] Of course, Gibson didn’t claim that we have direct perception of all of a thing’s physical properties (1972, p. 239); we are focusing on direct perception of metric size and shape at an appropriate scale.
[22] How do perceivers come to judge the metric sizes of things and to believe that things which look small in the distance are not actually physically smaller than when nearby? Granrud (2012) studied the development of this ability in children and found that they learn to adopt a cognitive strategy (that things look smaller in the distance) when making judgments of objective (physical, metric) size as opposed to apparent size. Accordingly, perception presents objects under an aspect as phenomenally diminishing in direct relation to phenomenal distance, and perceivers learn to take distance into account when making judgments of objective size based on such appearances.
[23] Gibson (1950, 1959) advocated a new psychophysics to reveal stimulus variables for perception as opposed to sensation (as in traditional psychophysics). He later found this psychophysics to be inadequate; even so, Lombardo is right that: “The ecological approach did not replace the psychophysical approach so much as absorb much of it into a broader, more fundamental context” (1987, p. 246). Gibson (1979, p. 149) said that he was “mistaken,” in 1950, to seek the stimulus for perception. He now viewed visual information not as a cause of perception (and hence, strictly speaking, not as a “stimulus”) but as an opportunity for sensing the environment (1979, pp. 56–57, 150). But a “psychophysics” of perception might still be useful: “what I should have meant by a ‘psychophysical’ theory of perception in 1950 and by perception as a ‘function of stimulation’ in the essay I wrote in 1959 (Gibson, 1959) was the hypothesis of a one-stage process for the perception of surface layout instead of a two-stage process of first perceiving flat forms and then interpreting the cues for depth” (1979, p. 150). He envisions a psychologically primitive (unanalyzable) process of resonance to information for the layout.
[24] Recently, Cormack (2017) studied binocular flow fields, including the detection of interocular velocity differences (between features across the two eyes moving toward or away from one another along the horizontal), as indicating increasing or diminished distance to the object or world.
[25] For discussion of how environmental regularities might be incorporated into a representational account of perception and yet avoid symbolic representations, explicit internal statements of rules, and constructed internal objects that are “seen,” see Kosslyn & Hatfield (1984) and Hatfield (1988).
[26] Van de Grind (1988, p. 170) grants that symbolic representations might be needed for higher cognitive tasks. 
[27] Above, I considered the functions of perceptual systems in relation to a task analysis, ascribing them the function of producing perceptions that can guide action, whether perceptions of the surface layout, the changes in things, or their affordances. That is all “teleofunctional” need mean here. On this conception, Gibson (1966b, 1979) provided task analyses, just as did Marr 1982 (his level one “computational” analysis); they differed over how to conceive the task of perception (as direct acquaintance with the mind-independent physical environment and its affordances, or as subject-relative representation of that environment).
[28] Frost & Sun (2003) find physiological evidence for looming detectors in pigeon. They speak of posited concentric rings of motion detectors as “signaling” symmetrical image expansion and as helping to (neurally) “compute” time to collision; and of postsynaptic potentials “representing” angular velocity or angular size (p. 30). They speak of “information” about time to collision and subtask outputs of information. The relation to Gibsonian information is complex. I have paired “content” with “information” in relation to subprocesses; “content” is not a technical term for Gibson but he did believe in it – or in a connate notion of “meaning” (1979, p. 33), applied to the layout and to affordances (I of course expand on his intended usage). Finally, on this analysis, optic flow detectors needn’t be seen as extracting a 2-D flow-image from which 3-D motion is inferred: neural computation can be construed as information combination, not inference, and detected optical expansion need not be viewed as a coherent image.
[29] Gibson resisted decomposing the response to optic flow. On my reading, he remained focused on personal-level psychological descriptions: information in the changing optic array (optic flow) is simply “picked up”; the evidence is that organisms are able to guide their action effectively. But he acknowledged that there is a further story involving receptor cells and patterns of stimulation, as found in the “pick up process”: “the action of the nervous system is conceived as a resonating to the stimulus information” (1966b, p. 271); we need to know more “about the inner, central nervous resonance to selected inputs” (p. 268); optical stimulus information exists at an observation point without a perceiver being there, but: “Actual stimulation depends on the presence of photoreceptors” (1979, p. 53); “In order to stimulate a photoreceptor, that is, to excite it and make it ‘fire,’ light energy must be absorbed by it” (p. 52). Gibson distinguished passive receptors from active perceptual systems that respond to stimulus information, but he also allowed that physiological receptors must be collected together to form the system that resonates to such information (1966b, p. 41; 1979, pp. 52–53; see also 1966b, pp. 5, 251, 253; 1979, p. 251).
[30] Van de Grind (1988, p. 168) locates “bad” talk of codes-and-information in conventional codes, arguing that his approach is based on a task analysis that examines natural capacities and so is not merely conventional: “In ‘mediated’ theories of perception one or more of the postulated stages involve symbol structures (i.e. internal representations), where the nature of the symbols, of the internal codes, is in principle arbitrary, that is, a matter of convention. In ‘direct’ theories there may be more stages or not, but the meaning of the internal activity lies in the behavioural action as a whole. There the code is not arbitrary (conventional) but specific to the problem and not to be understood in separation from the functional aspects of the whole situation.” Here he in effect adds a third option, between no representational codes and arbitrary symbolic codes: the option of imputing representational codes based in a task analysis of a perceptual system as a natural system. (Also, many symbolists would not agree that their codes are arbitrary, but that is not our problem.)
[31] This section argues that a Gibsonian might accept subpersonal representations in an analysis of pickup mechanisms without the sky falling. This can be done in a dynamic system such as that of van de Grind. Of course, many dynamic-systems fans sympathetic to Gibson eschew representation (e.g., Chemero 2009; Michaels & Palatinus 2014). My aim is not to exclude non-representational analyses of information pickup, but to portray a viable representationalist alternative. Concerning Gibson and (noncognitive or nonconceptual) representations, see Epstein (1993); Marr (1982, ch. 1); and Shapiro (2011, ch. 2). For arguments favoring representationalist construals of (at least some) dynamic systems, Shapiro (2011, ch. 5), and the literature cited there; in the Gestalt tradition, van Leeuwen (2015). As a note for the curious: the two contexts for representation addressed herein (personal-level or whole organism, and subpersonal functionally posited systems) share in common the instantiation of nonconceptual content.



Pièces mécaniques en chocolat - Lucca