by Corry Shores
[Search Blog Here. Index tabs are found at the bottom of the left column.]
[The following is summary. All boldface and bracketed commentary are my own. I apologize in advance for my typos. Proofreading is incomplete.]
Mark Bickhard and D. Richie
On the Nature of Representation:
A Case Study of James Gibson’s Theory of Perception
1.2 A Historical Summary of Gibson’s Theory
Very brief summary:
The important information about the world we perceive is not something our minds place into our perceptions by organizing and processing them one way or another. Rather, that information is built into the world (the “ecology”) itself and especially in the patterns, structures, and ways we perceive it when we actively interact with it. And we directly, without further processing, discern these significances in our perceived world when our perceptual interactions tell us the possible uses of (or potential further interactions with) the perceived things, which are called their “affordances”.
Gibson rejected two predominate views of perception of his time and proposed a new theory. The first predominate view are the sensation-based theories (of for example Berkeley, Müller, and Helmholtz) which say that our eyes directly receive and encode our fragmented visual data and secondarily we construct full perceptions on their basis by applying to them processes of memory, inference, and judgment. The other view is the Gestalt one, which sees the process of perception as a relatively spontaneous sensory organization that also involves some reconstructive work on the part of the perceiver. But Gibson’s own scientific studies showed that humans and animals react to their environment in a way that is too accurate and immediate for there to be additional acts of processing of the data, as in these two theories. Instead of the important information being encoded into the perceptual data by means of perceptual and mental processes, Gibson instead came to hold that the important information about the environment was fully given in the sensory data already and directly discerned without further processing of it. His work with motion parallax illustrates his thinking. (Motion parallax is the visual experience that we have when we are moving, and the things further away from us pass through our field of vision slower compared with nearer things.) The two existing models would say that on the basis of a static impression of what is given to our vision while we remain passive observers, we process that data to obtain knowledge of the properties of depth in the scene before us. However, motion parallax shows us that we are not passive observers, because we actively make decisions about how we position ourselves in the world, and in that way we in fact interact with our surroundings. And also, what we perceive is not a static image, because it requires that we see a flow of information and we recognize certain patterns in it. But it was not so clear in his early theories that he can avoid what he called the homunculus problem, which plagued the other theories. The idea is that because an interpretative sort of process was needed to “encode” the raw data with other important significances, like depth relations, there needed to be an internal agency that receives that data and interprets it, like a miniature human living in our brain. This leads to an infinite regress, because that homunculus would need some internal interpretative agent of its own as well, and so on. Gibson, thus, needs to explain the nature of the discerned significant information and the way it is obtained in a manner that shows how it does not involve an interpretative component. This danger arises on account of his notion that schematic perception is based on literal perception. Our literal perceptions give us data about the physical spatial properties of things in our perceived world, while our schematic perceptions tell us about the perceived things’ potential uses and significances. Schematic perception is in some sense obtained from literal perceptions secondarily, and this leaves open the possibility that Gibson’s theory falls to the homunculus problem (for, it could be that there needs to be an additional process that “reads” the literal perceptions to interpret their schematic significances). What he says instead is that humans and animals, by perceiving the physical features of things, in that same act thereby perceive their potential uses or interactive possibilities, called their affordances, of that perceived thing. Thus there is no ‘encoding’ or processing to obtain knowledge of the perceived thing’s significance; it is perceived and recognized directly when seeing its physical features. Given that this interactive element with the environment directly contains the important information about it, Gibson moved the locus of perception away from the passive perceiver’s internal workings and relocated it in the surrounding “ecology” of interwoven environmental elements that are also interactively interwoven with the observer. [For this reason, it is called “ecological psychology,” as Gibson argued that] perception is a “process that can only be understood in terms of its natural ecology” (Bickhard & Richie 10).
Bickhard and Richie (henceforth written as BR) will discuss James Gibson’s theory of perception by first examining its historic context and then its conceptual development (BR 8).
BR begin by noting the basic context of Gibson’s original studies:
Gibson (1950) points out that the study of perception had long been dominated by the problem of how the mind can generate our full experienced perceptual knowledge from the inadequate data provided by the senses, with vision and the eyes always the primary focus. The major approaches to this problem were based on the works of Berkeley (1709/1922), Müller (1838/1948), and Helmholtz (1896/1952), who proposed that the eyes directly receive and encode certain basic sensations, such as patches of color, lines, points, and so on, and that full visual perceptions are then constructed on the basis of such sensations through various processes of comparisons with memory, inferences based on cues within the sensations, and, ultimately, judgments concerning the nature of the external stimulus.
Although there were disagreements on the nature of the sensations and of their processing, “all such models, including a slight variant in which the retinal image served in the role of sensations, assume that perceptions must be generated out of primitive sensations or retinal images. They assume that the senses receive fragmented or incomplete information about the world that must be enriched by mental processing (Gibson & Gibson, 1955)” (BR 9a).
Gestaltists “objected to this approach,” because they thought that “the sensory elements seemed impossible to specify” and also that this approach at best tells how we make judgements about the world but does not explain how we actually see the world [in a more encompassing sense] (9). Rather, “Gestaltists argued that ‘experience is not reducible to elements or additive units’ and proposed instead that the process of perception ‘was one of a relatively spontaneous sensory organization’ (Gibson, 1950, p. 22)” (9). This notion of sensory organization applied well to the perception of form, but not so well to the perception of space. And in both cases it was difficult to specify [what the nature of the sensory organization is]. What interested Gibson with regard to the Gestaltists was that they “formulated genuinely relevant problems for space perception, problems concerning the characteristics of the actual experienced visual world rather than the flat geometric visual field (Gibson, 1950, p. 23)” (9).
With these two theories in mind, Gibson conducted his own experiments in depth perception during the second World War (9). What he found was that “depth perception was more accurate than could be explained by any model based on depth cues” (9). This meant specifically that the sensation models failed but also the Gestalt theories proved inadequate [for some reason] as well (9).
In 1950, Gibson writes The Perception of the Visual World, and here he goes beyond both theories. BR assesses the role of these alternative theories in this way:
From the Gestaltists, he accepted and adapted the idea that the most basic problems of visual perception were those regarding the experienced three-dimensional visual world, not the flat geometric visual field, but he rejected the proposed process of sensory organization. From the sensation-based approaches, he accepted very little, neither their basic problems nor their basic solutions.
[I am not certain, but perhaps we can say the following about Gibson’s critique. He will say that humans and animals react to the spatial environment in a way that shows they have a very precise understanding of its spatial features. This means that the two theories fail. The sensation-based approaches perhaps do not explain this, because this theory might say that people’s and animal’s understanding of the spatial environment is based on just what they sense. But too little is sensed to support such a detailed knowledge of the environment. The Gestalt approach does not work, because this would say that the mind constructs a lot of its knowledge of the environment artificially. But were that so, there would be more errors in that construction than are actually there. Let me quote, as that was just a guess.]
Gibson argued that people and animals “appear to react to the spatial environment with an accuracy and precision too great for any known theory of space perception to be able to explain. ... If the solid visual world is a contribution of the mind, if the mind constructs the world for itself, where do the data for this construction come from, and why does it agree so well with the environment in which we actually move and get about” (p. 14). This basic rejection of mental constructivism, of mental processing was one of the most fundamental moves in the development of Gibson’s own theory. Consistent with this rejection, Gibson also rejected the premise that made such processing necessary and the particular distinctions and processes by which it was presumed to occur.
(BR 9-10, citing Gibson 1950)
Gibson especially “rejected the basic premise that the data available to the senses were inadequate to perception” (10). [I am not sure, but this notion might work against both theories. The Gestaltists seem to be saying that there are additional constructions and organizations on the basis of what is given, and the sensation-based approaches discuss certain unconscious cognitive processes that formulate inferences and judgements on the basis of what is given. But if what we perceive is already adequate, none of this additional work is needed to develop a sufficient understanding of the world around us.]
In particular, and most fundamentally, Gibson rejected the basic premise that the data available to the senses were inadequate to perception: “Even complex perceptual qualities must have stimuli” (p. 8); “If the total stimulation contains all that is needed to account for visual perception, the hypothesis of sensory organization is unnecessary” (p. 25). Clearly, if the total stimulation contains all that is necessary to account for visual perception, then the (unconscious) inferences, comparisons with memory, and judgments – the mental processing – of the sensation-based models are also unnecessary. If we ask the right question, Gibson suggests, if we ask about the experienced visual world based on surfaces and edges, rather than about the flat geometric visual field, then we find that the information available to the visual senses is sufficient to perception, and information enhancement via mental processing is a superfluous and flawed postulate.
(BR 10, citing Gibson 1950)
Because Gibson rejected this sense of mental processing as involving enhancements, he thereby also rejected “the classical distinction between sensations and perceptions;” for, “that distinction is based on the assumptions that sensations are informationally impoverished and that mental processing enriches them into perceptions” (10). [In other words, according to this view that Gibson rejects, we have raw sensations that are inadequate possibly because they are fragmented and/or disorganized, and thus perception is the process by which these sensations are made adequate perhaps by organizing them and/or by completing them where they leave informational gaps.]
Sensation-based models often see the perceiver as passively and statically receiving sensations on the basis of which she forms perceptions. Gibson rejects this idea, because human perceivers are very active while perceiving, making spontaneous decisions like changing where to look and how to orient themselves in their environment in order to perceive it better. Gibson also argued that perception is a “process that can only be understood in terms of its natural ecology” (BR 10). [Perhaps by this is meant that perception is always bound up with the conditions of the environment and one’s interactions with it, but I am not sure.] One way that Gibson supported this notion was by noting ways that changes the perceiver makes in their motion within their setting can enhance their perception, which can in fact even “modify the retinal images in a quite specific way”, as when for example certain physical movements provide “powerful information for depth perception in the form of motion parallax” (10).
[I am not very familiar with this notion of motion parallax. I found this helpful diagram.
Motion parallax seems to be that very familiar phenomenon where when we are moving, objects in the distance seem to pass through our field of vision slower in comparison to things nearer to us. We might normally notice this when looking out of a train or automobile window. Until I read the source text by Gibson, I will not know what to say about this. But perhaps we might note at least that here depth perception is attained by moving around interactively in an environment, rather than simply taking in some sensory information passively, processing it, and then discerning the depth afterward.]
Gibson had yet another strong argument against the sensation-based models of perception, namely, the “homunculus problem” (10). To understand this issue, we should first take note of “retinal-image theories,” which say that it is necessary for people to process the stimulation on the eye’s retina. Gibson argues against this. He first observes that this view thinks that an image forms on the retina like an image projected onto a screen, and thus “the retinal image is something to be seen” (BR 11, quoting from Gibson, 1979, p. 60). [The idea here is that there is a sort of second act of seeing, that is, seeing the image on the “screen” of the eye.] Gibson considers this then a matter of a homunculus problem, because it is, as he calls it, “the little man in the brain theory” of the retinal image (BR 11, citing Gibson 1966, p. 226), “which conceives the eye as a camera at the end of a nerve cable that transmits the image to the brain. Then there has to be a little man, a homunculus, seated in the brain who looks at the physiological image. The little man would have to have an eye to see it with, of course, a little eye with a little retinal image connected to a little brain, and so we have explained nothing by this theory” (11 again from quotation). In fact, this theory only makes the matter worse, because it entails an endless series of little perceivers inside little perceivers (11). As BR explain, there other versions of this argument, but they all have in common “that something, or someone, must ultimately do the perceiving, and that is what was to be accounted for in the first place” (BR 20). They say that this homunculus problem is found in “any form of inputs-followed-by-processing-followed-by-perception model” (11).
Sensation-based models posit constructions by means of perceptual processes in order to account for “the problem of how full perceptions are derived from impoverished sense data” (11). But Gibson’s “assertion that the total stimulation is informationally adequate to perception” rejects the assumptions underlying that problem (11). “Gibson continued to develop his arguments against sensation-processing and other input-processing models” (11).
Gibson, seeing the shortcomings of sensation-based and Gestalt models of perception, offered his own model that BR describe as “an ecological direct-encoding model” (11). [The notion of an ecological model of perception is not very well defined for me yet, but perhaps that becomes clearer as we continue. It might mean a model in which the perceiver takes an active role in interacting with the perceived world while perceiving it. In this sense it might be something like Merleau-Ponty’s notion of how the perceiver is integrated with the world they perceive, forming one flesh. The fact that it is encoded seems to mean that the information is still discerned by placing it into what is perceived, but this is somehow done directly. But I am not sure. From what is said later, the notion of direct encoding might be related to the idea of resonance. So maybe our mind encodes the information into the sense data by resonating with the information that is already in a sense encoded in the way the data is given to us. But that is wild guessing on my part.] BR continue,
Gibson rejected the sensation-based conception of the perceiver as a passive individual confronting a flat visual field in favor of an active perceiver confronting an ecologically structured visual world – thus, | an ecological model. He also rejected both the mental constructivism of the sensation-based models and the sensory organization of the Gestaltists in favor of a direct correspondence between stimulation and perception thus, a direct encoding model.
BR then note how “The direct encoding aspect of Gibson’s 1950 model was both a methodological move and a theoretical move” (12). The methodological component is his locating the basic problem of perception as being the problem of “establishing an empirical correspondence between the stimulus and its conscious resultant,” and this means that he proposes a sort of ecological psychophysics (BR 12, quoting Gibson 1950, p. 52). The theoretical component lies in the fact that he “rejected any intermediate processing of encoded sensations between the stimulation and the perception and in his corresponding rejection of the sensation-perception distinction” (12).
BR acknowledge that “It is not entirely clear that Gibson would have agreed with the ‘encoding’ part of our designation of his model as an ‘ecological direct-encoding’, especially in his late career” (BR 12). [Our summaries so far have skipped over BR’s own way of using Gibson’s thinking.] But, his 1950 model “seems committed to some form of a direct encoding model” (12).
If we adopt an encoding model, then we then have to explain how these encodings occur (12). As we will see, his “conceptualization of an ecologically active perceiver contains the germ of his later answers to that question and, we argue, the germ of interactive insights that allowed him to largely transcend the encoding approach altogether” (12). BR will now discuss the later development of his model.
The later model develops the “internal implications” of the 1950 model (12).
In the 1950 model was a sort of retinal-image-based model of perception, and “He described his psychophysics program as involving a ‘jump from the retinal image directly to the perceptual experience’ (p. 51)” (BR p. 12, qtg. Gibson 1950).
But as Gibson came more to an “ecological emphasis on the importance of the active perceiver,” this retinal-image focus proved inadequate (12d). For, “The retinal image of an active perceiver changes too much, too fast, and too continuously, in contrast with relatively stable perception, to be the primary locus of perception” (BR 13a). Gibson writes, “The active observer [however] gets invariant perception despite varying sensations” (BR p. 13, qtg. Gibson 1966, p. 3, bracketed insertion is BR’s).
So Gibson needed to find a “different locus of perception information” (13). It would need on the one hand to remain stable like how our perceptions are, while at the same time be “adequate to those perceptions” (13). [I am not sure what is meant by them being adequate to the perceptions. Perhaps the idea is that his model would need to juggle both the fact that sensory information is highly variable and thus not similar to perceptions, which are stable, while at the same time needing to somehow have this stability that perceptions have. So if we say that the sensory information is directly perceived but is variable, then it is not adequate to the perceptions, which are less variable. I am not certain however.] [The notion of a “perceptual locus” is important here, but I am not entirely sure I grasp what is meant by it. It seems to me that BR are saying that for Gibson, the locus of perception is actually somehow in the environment. But I am not exactly sure what is meant by that yet. They will refer again to motion parallax, and they write, “Motion parallax is a phenomenon of the structure of the ambient light through which the eye moves. The clear suggestion is that the broader spatial and temporal patterns in the ambient light might well be the actual locus of visual perception.” So the idea here might be the following: the structures by which depth is discerned are found not in the way the visual information is processed in the perceiver’s mind but rather it is located within the structures of the visual data itself as it is already in its raw givenness. So contained within the visual data hitting our eyes when we pass through the fields of light beams found in the space along our train ride, there is already the far away items “moving” slowly relative to the faster moving things nearer and nearer to us. Gibson might then be saying we thus perceive the depth directly, because that depth is already built into the way the light beams are structured in their patterns of givenness. Thus, we have a direct perception of depth on the basis of whatever visual data we get in whatever way it is given, without needed to process it.]
A different locus of perceptual information was required, one that maintained a stability comparable to that of perceptions and one that was adequate to those perceptions. A new perceptual locus was required by Gibson’s recognition of the importance of the active perceiver; such a locus was suggested by that same recognition. Gibson’s original emphasis on the active perceiver stemmed in part from the motion parallax information concerning depth that was thereby derived. Motion parallax is a phenomenon of the structure of the ambient light through which the eye moves. The clear suggestion is that the broader spatial and temporal patterns in the ambient light might well be the actual locus of visual perception.1 Certainly, on the one hand, there is no information available in the retinal image that is not available in the ambient light, and, on the other hand, it is difficult to conceive what alternative external locus for visual perception might be possible. Furthermore, very encouraging success was obtained in investigating the information that was in fact available in the ambient light. Correspondingly, “In my book, The Perception of the Visual World (1950), I took the retinal image to be the stimulus for an eye. In this book I will assume that it is only the stimulus for a retina and that ambient light is the stimulus for the visual system” (1966, p. 155).
[Endnote 1 on page 85 (quoting):
1. Such a shift to patterns in the ambient light as the locus of perception is clearly prefigured by his 1950 point that patterns of stimulation could themselves be stimuli (p. 9), even though at that time he was referring to retinal patterns. The shift is also consistent with his general ecological emphasis, but neither of these points is sufficient to force that shift – the active observer is sufficient.]
RB note two revisions that are called for in light of Gibson’s discovery. The first revision shifts the “postulated locus of visual perception from the retinal image to the ambient light” (13). [I might not follow the second revision. It seems to be that as a result of the first revision, we now need to change our view of the perceiver as someone who simply interprets the visual data given on the retina to instead be an active participant in the perceptual process by making spatial modifications in order to find patterns in the resulting changes in the visual data. Let me quote.]
Thus, consideration of the fact and necessity of the active perceiver forced a shift in the postulated locus of visual perception from the retinal image to the ambient light. Consideration of the ambient light as the locus of perception forced, in its turn, a reciprocal revision of the conception of the perceiver. The logic of the second revision derives from the fact that such broader spatial and temporal patterns in the ambient light cannot simply be sought by the visual system, then, when found, statically, retinally perceived. They are, by definition, too big for that. They must be scanned, sampled, or otherwise interacted with in such a way as to detect and identify - to pick up – an encounter with a discriminable pattern.
The detection and differentiation of such a broader pattern, a variant or invariant in the ambient light-the pickup of such information – is intrinsically interactive. The active perceiver of 1950 had to become a truly interactive perceiver: |
There is a loop from response to stimulus to response again (1966, p. 31).
An explanation of constant perception ... should be sought in the neural loop of an active perceptual system that includes the adjustments of the perceptual organ. Instead of supposing that the brain constructs or computes the objective information from a kaleidoscopic inflow of sensations, we may suppose that the orienting of the organs of perception is governed by the brain so that the whole system of input and output resonates to the external information, (1966, p. 5).
The process of pick up is postulated to depend on the input-output loop of a perceptual system (1979, p. 250)
The process is circular, not a one way transmission (1979, p. 61)
The course of the whole interaction can be critical. It is the course of the interaction by the visual system, for example, the scanning, both input and output and the relationships between them, that differentiates the pattern interacted with; it is not the ‘final’, static, retinal image that ‘completes’ the interaction that picks up such a pattern, nor even the ‘succession of images’ or, better, the flow of retinal stimulation that accompanies the interaction. Retinal stimulation is relegated to the input side of an overall interactive visual system that engages in such interactions and discriminates such patterns. It is the pattern of the interaction that differentiates and, thus, identifies the pattern interacted with; it is not any piece or component of the interaction.
(BR pp. 13-14, block qtg. Gibson)
[The next point reminds me very strongly of what Merleau-Ponty writes in section 1.2.1 of The Structure of Behavior. He describes the dynamic process of responding to a stimulus. Humans and animals do not know at the very start of a stimulus the correct response to it, because what needs to be recognized in the stimulus is a pattern that unfolds over time. And, while that pattern is unfolding, the responding creature modifies its receptivity in real time so to better sense the stimulus and respond appropriately. One example is how the ear of a cat responds to different sorts of touches:
Five different reflex responses can be obtained by stimulating the ear of a cat depending on the structure of the excitant employed. The pinna of the ear flattens out when it is bent, but responds to tickling with a few rapid twitches. The character of the response is completely modified depending on the form of electrical excitation (faradic or galvanic) or its strength; for example, weak strengths evoke rhythmic responses; strong ones evoke tonic reflexes. [...] (Sherrington and Miller). [qtd in Merleau-Ponty English translation p.11 / French p.9]
He also seems to have a view of the perceiver not being passive, as it interacts with the stimuli in order to perceive it in such a way as to respond to it properly.
The organism cannot properly be compared to a keyboard on which the external stimuli would play and in which their proper form would be delineated for the simple reason that the organism contributes to the constitution of that form. (Merleau-Ponty 13 / 11)
And he has the example of holding an animal in an instrument and adjusting one’s hold in response to the creature’s movements. The idea here seems to be that if the animal changed its body shape so to escape the instrument, then we lose the ability to feel its movements. However, if we sense its body changing its shape and immediately respond by changing our hold on it, then we can continue sensing it. In other words, perception involves self-modification in immediate interactive response to what we are perceiving.
When my hand follows each effort of a struggling animal while holding an instrument for capturing it, it is clear that each of my movements responds to an external stimulation; but it is also clear that these stimulations could not be received without the movements by which I expose my receptors to their influence. “... The properties of the object and the intentions of the subject ... are not only intermingled; they also constitute a new whole.” (Merleau Ponty 13 / 11; the quotation is cited as “Weizsäcker, Reflexgesetze, p.45. “L’organisme est, dit Weizsäcker, Reizgestalter.” [Note: Reizgestalter is misspelled as Reizgestaller in the English translation.])
And he illustrates the real-time modifications to receptivity with how telephones seemed to work at that time. Apparently you dialed the receiving person’s name. After dialing the first letter, the connecting station then becomes sensitive to only those sets of letters that it knows can come after that first letter, and so on. (But I am not sure exactly how these phones worked.)
The model of the automatic telephone appears more satisfactory. Here indeed we find an apparatus which itself elaborates the stimuli. | In virtue of the devices installed in the automatic central, the same external action will have a variable effect according to the context of the preceding and following actions. An "O" marked on the automatic dial will have a different value depending on whether it comes at the beginning, as when I dial the exchange "Oberkampf," for example, or second, as in dialing "Botzaris." Here, as in the organism, it can be said that the excitant — that which puts the apparatus in operation and determines the nature of its responses — is not a sum of partial stimuli, because a sum is indifferent to the order of its factors; rather it is a constellation, an order, a whole, which gives its momentary meaning to each of the local excitations. The manipulation “B” always has the same immediate effect, but it exercises different functions at the automatic central depending on whether it precedes or follows the manipulation “O,” just as the same painted panel takes on two qualitatively distinct aspects depending on whether I see a blue disc on a rose-colored ground or, on the contrary, a rose-colored ring in the middle of which would appear a blue ground. In the simple case of an automatic telephone constructed for a limited number of manipulations, or in that of an elementary reflex, the central organization of the excitations can itself be conceived as a functioning of pre-established devices: the first manipulation would have the effect of making accessible to subsequent ones only a certain keyboard where the latter would be registered.
(Merleau-Ponty 13|14 / 12)
Gibson’s point of course is not identical to Merleau-Ponty’s, but let us note what seems to be two important similarities. In both cases, what is being perceived is something dynamic, and its unity is to be understood in terms of a pattern of variation. The second similarity is the interactive nature of the perception. The perceiver cannot simply remain in the same mode of receptivity. Rather, they need to adjust or modify themselves (in relation to their environment in general or to the external stimulus specifically) in one way or another in order to properly perceive the important patterns of the stimulus’ dynamic variations.]
BR next discuss the parallax example in terms of information. The light patterns are the information itself, found in the field of ambient light. BR also characterize the perceptive act by which the depth is discerned as involving information-extracting interactions as a means of picking up information. [But I would suppose this is not a matter of processing the information but rather of picking out the information already given immediately.] But Gibson does not mean for “information” to have its normal sense of “knowledge communicated to a receiver,” because Gibson does not want to imply necessarily that the information needs to be encoded and communicated to the perceiver [rather than being completely apparent from the beginning and immediately available as such to the perceiver (without further ‘encoding’ or ‘decoding’)] (BR 14).
Gibson does still think that retinal stimulation plays a role in visual perception. His emphasis however is on the nature of that role they play [I am not entirely certain, but it seems Gibson thinks the role is the following (and then I will quote so you can check). The role of the retinal stimulation is to provide the information in its physiological form, and perhaps we are to think of it as nervous signals. But Gibson emphasizes that the information about depth for example does not need to be acquired by further processing that visual information; for, it is already built into the structure of that information and it can be directly (without mediation) discerned by the perceiver.]
Gibson was also well aware that retinal stimulation does occur, that it plays a central role in visual perception, and that it is involved in (interactive) processes. The issue is the nature of that involvement: “The inputs of the receptors have to be processed, of course, because they in themselves do not specify anything more than the anatomical units that are triggered” (1979, p. 251). Information, however, “is not something that has to be processed” (1979), p. 251). “Information is conceivable as available in the ambient energy flux, not as signals in a bundle of nerve fibers” (1979, p. 263). Information is extracted by the interactions of sensory systems, not | encoded and transmitted by sensory organs. The eye and its stimulations participate in information-extracting patterns of interactions; they do not encode that information.
Gibson’s interactive theories of perception involve the criticism of the notion of encoding and decoding in perception (BR 15).
The information in ambient light [with regard to parallax motion] does not need to be encoded, but there still needs to be a “process of pickup,” which he explained using the metaphor of resonance:
The perceiver interactively resonates with the available information (for example, 1966, p. 5; 1979, p. 246). Consistent with this suggestive metaphor, he also referred to the process of becoming able to extract information, of learning to resonate to available information, with a metaphor of “tuning.”
But resonance is not the only way “energy patterns can be picked up without intermediate enhancement of encoded information” (15). Another problem with these metaphors is that “resonance requires periodicities in patterns to resonate to, and those are not necessarily available in information to be perceived” (15). A third problem is that “even if such periodicities were available, it is neither at all clear what it is about the interactive loop that would resonate to them nor how it would do so” (16). And a final problem with these metaphors, and also the most important one, is that “that which resonates generally resonates at the same (or a directly related) frequency as that which is resonated to. The resonant frequency is a copy, a duplicate, of the original frequency. Such vestiges of picture, of image, of encoding conceptualizations are regretfully distortive of Gibson’s basic interactive insight in his concept of information extraction. The pattern of an interaction need not have any particular structural correspondence whatsoever with the pattern of ambient light that it differentiates” (15).
[So this metaphor of resonance is not entirely helpful for understanding the process of picking up information by means of interactive perceptions of the patterns in the environment’s dynamics. This also means that, without other explanations, we might have difficulty conceiving how this pick up process works.] But despite these problems with the metaphors, “the basic direction of the evolution of Gibson’s theory seems clear” BR explain (15). And in fact, that development continued even after his discovery of “interactive information extraction” (15). But Gibson’s model could involve the homunculus problem, [because the nature of the extraction has not been specified] so he needed to make a further step in the theory’s development, and that step “involved the problem of meaningful perception” (15).
Gibson already in his 1950 book The Perception of the Visual World worked with a notion of meaningful perception. There he “made a distinction between ‘The perception of the substantial or spatial world and ... the perception of the world of useful and significant things to which we ordinarily attend’ (p. 10 italics omitted)” (BR 15). Gibson’s term for the perception of the substantial or spatial world is literal perception, and the perception of the world of useful and significant things to which we ordinarily attend is called schematic perception. BR explain that “Schematic perception was presumed to be based on literal perception because literal perception ‘provides the fundamental repertory of impressions for an experience’ (p. 10), and the two forms of perception were presumed to have importantly different properties. Meanings were presumed to be attached to, and detachable from, the spatial impressions of literal perception” (BR 16).
BR then explain how this relates to the homunculus problem. For there to be meaningful perception, it would seem to require a homunculus to receive the literal spatial impressions and interpret them as having their appropriate meanings (16). And thus “Literal spatial impressions must be enhanced, presumably via some kind of processing with meanings” (16).
BR trace Gibson’s solution to this problem to beginning steps in his work of 1950, where he tied the usefulness of objects to their spatial features [and perhaps thus also our literal perceptions were tied to our schematic ones.] Gibson calls this squeezability: “He recognized [...] that ‘squeezableness is something which seems to be located in the object, not in the hand.... Visual objects appear to have soaked up such qualities and to be fairly saturated with them, the use of the object and the shape of the object being almost indistinguishable’ (pp. 203, 204). The idea that needs to be avoided still is that “the perception of the functional nature is dependent on the perception of the spatial nature” (16).
BR then have us look at Gibson’s model in a light that does not necessarily lead to this idea. We know already that when [for example in the case of parallax motion] we directly perceive the information [about depth], what we perceive are patterns that result from interactions. And furthermore, the information that we obtain indicates “potentialities for further actions and interactions” (BR 16). In other words, “what are most directly perceived are functional potentialities, potential usefulnesses” (16). [So when for example we interact with the environment by moving around it in order to obtain data that directly tells us of its spatial feature of depth, what that tells us are the different sorts of spatial ways that we may further interact with that space by moving through it in all its available dimensions.] Thus these “patterns of interactions [...] are simply functional indicators” (BR 16). [BR continue in the endnote to this passage: “From this perspective, in fact, the spatial is subsidiary to the functional. Surfaces, objects, and the like are constructed as patterns of potential interactions, including further perceptual interactions, that may be indicated by particular perceptual interactions, that is that may be perceived. Such construction of the physical and spatial out of the functional is in the general spirit of Piaget” (BR, endnote 2, p.85).] [I am not exactly sure what is meant here by “functional indicators”. Perhaps they are indications of ways that certain interactions with the environment can produce certain types of possible results. So for example, knowledge of how far a mountain is away from us, along with its relative height in comparison with its surroundings, can indicate the sorts of views we might have were we to climb it and the approximate amounts of time and effort it would take to accomplish that.]
What he previously called squeezability he later refines into his notion of affordance. [An affordance seems to be the directly apparent uses of things we perceive. So in the same act by which we observe the physical features of something, we thereby perceive its usability for certain purposes. But this usability is observed directly, because it is directly evident in the thing’s physical features.]
Such an imbuing of perception with direct, functional, ecological meaning, already hinted at in his 1950s discussion of squeezability, yield Gibson's concept of affordance. “The affordance of anything is a specific combination of the properties of its substance and its surfaces taken with reference to an animal” (1977, p. 67, italics omitted). Affordances are those things the environment “offers the animal, what it provides or furnishes either for good or ill" (1979, p. 127).3 And such affordances are intrinsic to perception:4
The composition and layout of surfaces constitute what afford ... to perceive them is to perceive what they [surfaces] afford ... it implies that the “values” and “meanings” of things in the environment can be directly perceived (1979, p. 127).
The perceiving of an affordance is not a process of perceiving a value-free physical object to which meaning is somehow added in a way that no | one has yet been able to agree upon; it is a process of perceiving a value-rich ecological object (1979, p. 140).
[Endnotes 3 and 4 from page 85 (qtg.):
3. Affordances, of course, are therefore “relative to the animal. They are unique for that animal. They are not just abstract physical properties” (Gibson, 1979, p. 127). “Knee-high [therefore affording the potentiality of sitting on] for a child is not the same as knee-high for an adult” (Gibson, 1979, p. 128). Horizontal support for water bugs is different than for heavy terrestrial animals (Gibson, 1979, p. 127).
4. Gibson’s discussion, however, still suggests too much independence of the spatial from the functional; there is an incomplete recognition of the construction of physical and spatial representation out of functional representation. (Such construction would be a part of Gibson’s tuning, not his information extraction.) Gibson still wants to go “from surfaces to affordances,” he does so by having “the composition and layout of surfaces constitute what they afford” (1979, p. 127), but such constitution still leaves the question of what representation of a surface is as logically prior, though no longer temporally prior, to a representation of an affordance. Yet infants can perceive affordances without necessarily perceiving the surfaces, edges, and full objects that provide, or constitute, those affordances.]
BR’s final point seems to be that originally Gibson had a notion of ecological direct encoding, where we directly perceive the information because our minds somehow resonate with it [and in that way “encode” it in the sense of endowing its internal form with an informational value that was already there in its external form, but I am guessing]. But on account of there needing still to be some process encoding that information, it was abandoned later for this notion of affordance and interactive information extraction, where the meaning or information to be discerned in something perceived is given by interacting with it, which gives us its important physical features that thereby directly informs us of its possible uses. In other words, the significance of a perceived thing is its potential uses, and that is directly perceived by interaction with it and its environment rather than by some cognitive process whereby that information about its significance is decoded from our static perceptions of it. [Note, I might be missing the idea, because BR are using the term “encoding” rather than “decoding”. I am a little confused how it all works. Apparently according to some models, somehow at the site of sensations there is an encoding of information, and perhaps, but I am not sure, these models would have to say there later is a process that decodes this information (as if by a homunculus). The model is not very obvious to me yet. I wonder by the way if the encoding is anything like Lotze’s “local signs”.]
Interactive information extraction and affordances were the culminations of Gibson’s major moves away from his early ecological direct encoding. Although we later argue that those moves were nontrivially incomplete, nevertheless they transcended that early encoding model by constructing an intrinsically interactive mode of perception. Essentially, Gibson started with ecological direct encoding, then filled in the detection-differentiation-identification process, the process of ‘transducing’ the encodings, with so much interactive activity – extraction, resonance, pickup, affordance – so as to make it clear that whatever ultimate perceptual encoding, if any, occurred it was not primary nor necessary nor independent, but, rather, subsidiary to interactive extraction. Gibson’s basic insight was that it is possible to derive information about an environment from interactions with that environment without encoding anything from that environment.
Bickhard, Mark, & D. Richie. On the Nature of Representation: A Case Study of James Gibson’s Theory of Perception. New York [and other cities]: Praeger, 1983.
Bickhard and Richie cited a number of Gibson sources:
Gibson, J. J. The ecological approach to visual perception. Boston: Houghton Mifflin, 1979.
Gibson, J. J. The perception of the visual world. Boston: Houghton Mifflin, 1950.
Gibson, J. J., & Gibson, E. J. Perceptual learning: Differentiation or enrichment? Psychological Review, 1955, 62, 32–41.
Gibson, J. J. The senses considered as perceptual systems. Boston: Houghton Mifflin, 1966.
Or if otherwise noted:
Merleau-Ponty, Maurice. The Structure of Behavior. Transl. Alden L. Fisher. Boston: Beacon Press, 1963.
Merleau-Ponty, Maurice. La structure du comportement. Paris: Presses universitaires de France, 1942 / 1967.
Automobile parallax motion diagram:
Travis Schirner, ¨Mission Possible?¨