{"id":4821,"date":"2010-07-08T12:20:31","date_gmt":"2010-07-08T12:20:31","guid":{"rendered":"http:\/\/www.labri.fr\/perso\/barla\/blog\/?p=4821"},"modified":"2012-07-09T16:04:33","modified_gmt":"2012-07-09T16:04:33","slug":"seeing-and-visualizing-it%e2%80%99s-not-what-you-think","status":"publish","type":"post","link":"https:\/\/www.labri.fr\/perso\/barla\/blog\/?p=4821","title":{"rendered":"Seeing and Visualizing, it\u2019s not what you think"},"content":{"rendered":"<p id=\"top\" \/><em>Zenon Pylyshyn<\/em><\/p>\n<h2>The puzzle of seeing<\/h2>\n<ul>\n<li>Very little information is retained from one glance to another when the eyes move, or even when the eyes do not move but when the display disappears brie\ufb02y. &#8211; p18<\/li>\n<li>Information about a visual scene is not stored in pictorial form, but rather is stored in a form more like that of a description, which is characterized by variable grain and abstractness and is based on available concepts. &#8211; p36<\/li>\n<li>I propose a distinction between vision and cognition in order to carve nature at her joints, that is, to locate the components of the mind\/brain that have some principled boundaries or some principled constraints in their interactions with the rest of the mind. &#8211; p39<\/li>\n<li>In the case of the &#8220;perceived&#8221; size of an unfaliliar object, the object may not actually look different, depending on nearby size cues ; it may simply be judged to be a different size &#8211; p42<\/li>\n<\/ul>\n<h2>The independence of vision and cognition<\/h2>\n<ul>\n<li>The knowledge-dependent (or cognitively penetrable) stages include a preperceptual stage, wherein vision is directed at relevant places or objects in a scene, and a postperceptual stage, in which memory is accessed, and judgements are made about what is in the scene. &#8211; p63<\/li>\n<li>Perception appears to be rather resistant to rational in\ufb02uence. It is a remarkable fact about the perceptual illusions that knowing about them does not make them disappear &#8211; p64<\/li>\n<li>Principles of visual interpretation explain why it is that how you perceive one aspect of a visual scene determines how you perceive another aspect of the scene. When a percept of an ambiguous \ufb01gure reverses, a variety of properties appear to automatically change to maintain a coherent percept. p66<\/li>\n<li>The primary function of top-down pathways appears to be to allow attention to selectivelly sensitize or gate certain objects or regions of the visual \ufb01eld, as well as certain physical properties of the stimuli. p68<\/li>\n<li>One intriguing source of evidence that vision can be separated from cognition comes from the study of pathologies of brain function that demonstrate dissociations among various aspects of vision and cognition. p71<\/li>\n<li>What may be going on in the time it takes to reach perceptual closure may be simply the search for a locus at which to apply the independent visual process. This search, rather than the perceptual process itself, may thus be the process that is sensitive to collateral information. This is an important form of intervention, from my perspective, since it represents what is really a preperceptual stage during which the visual system is indeed directed by voluntary cognitive processes &#8211; though not in terms of the content of the percept, but in terms of the location at which the independent visual process is to be applied. &#8211; p80<\/li>\n<li>What the experts have tacitly learned is how to bring the independent visual system to bear at the right spatial location and what types of patterns to encode into memory &#8211; both of which are functions lying outside of the visual system itself. p86<\/li>\n<li>With regard to the genesis of visual expertise, it should be noted that there is no apriori reason why a postperceptual decision process might not, with time and repetition, become automatized and cognitively impenetrable, and therefore indistinguishable from the encapsulated visual system. p88<\/li>\n<\/ul>\n<h2>The architecture of the early vision system<\/h2>\n<ul>\n<li>The natural constraints view says that the visual system is so constructed (through evolution) that a certain sort of rigid interpretation will be the one automatically given (independent of knowledge of the particular scene ; indeed, even despite knowledge of the contrary) whenever it is possible, that is, whenever such a representation of the distal environment is consistent with properties of the proximal stimulus. p111<\/li>\n<li>Such constraints are quite different from the Gestalt laws (such as proximity and common fate) because they apply not to properties of the proximal stimulus, but to how such a stimulus is interpreted or used to construct a representation of the distal perceptual world. p112<\/li>\n<li>If the visual system is visually provided with an alternative &#8220;reason&#8221; for why dots are appearing and disappearing, then apparent motion is not seen. p115<\/li>\n<li>Many cases in which our visual system provides unambiguous (and usually veridical) percepts, despite the inherent ambiguity of the 2D image, can be explained without having to assume that the visual system draws inferences from general knowledge regarding what the 3D scene is likely to contain. Although the beliefs we come to have about what is in the scene invariably take into account what we know, the output of the early vision system, or to put it loosely, how things look, does not take into account such knowledge. The visual system is so constructed, presumably because its has become tuned over eons of evolutionary history, that the range of interpretations it is able to make is severely restricted. p121<\/li>\n<li>For the time being, we have the very general proposal that the visual system might be viewed as generating a set of one or more shape descriptors, which in many cases might be suf\ufb01cient (perhaps in concert with other visual information and presumably along with other information about the objects) to identify objects whose shape is stored in memory. p135<\/li>\n<li>In principle, the visual system could encode any property whose identi\ufb01cation does not require accessing general memory, and in particular that does not require inference from general knowledge. p136<\/li>\n<li>So far, the available evidence suggests that the visual system is not only cognitively impenetrable, but is also opaque to intermediate products of its processes. p139<\/li>\n<li>Much evidence suggests that there are severe restrictions on the \ufb02ow of information among the many subsystems involved in visual perception. For example there appears to be subsystems within vision with restricted access not only to general world knowledge but also to the outputs of sister systems. p143<\/li>\n<li>At the present time, the evidence is compatible with there being a single system that provides outputs separately to the motor-control functions and to the cognitive functions. p150<\/li>\n<li>The major difference between the information needed for the two types of functions is that motor control primarily requires quantitative, egocentrically calibrated spatial information, whereas the cognitive system is concerned more often with qualitative or categorical information, with relative locations, and with viewpoint-independent representations. p151<\/li>\n<\/ul>\n<h2>Focal attention<\/h2>\n<ul>\n<li>It is widely held that covert attentional movements precedes the overt movement of the eyes to certain locations in the scene, though such a covert movement may not be essential for an eye movement to take place. p168<\/li>\n<li>In addition to covertly controlling the locus of spatially localized attention, is has also generally been held that people can control the extent or breadth of the focus of attention. p168<\/li>\n<li>The process of individuating objects is separate and distinct from that of recognizing or encoding the properties of the objects. p174<\/li>\n<li>It seems as thought attention is required to bind properties together. This may be interpreted as suggesting that when features are detected, their location remains uncertain until the feature is attended. p177<\/li>\n<li>This is compatible with the view that individual objects as such are detected \ufb01rst, before any of their properties (including their locations) are encoded. p180<\/li>\n<li>The fact that visual information appears to be chunked according to individual visual objects suggests that encoding of visual information proceeds by objects, not by features. This in turn suggests that the attentional bottleneck operates over individual objects and that what happens early on in the visual pathway is that objects are individuated or selected. p182<\/li>\n<li>Objecthood endures certain kinds of location changes, and when it does, it carries object-based effects along with it. p188<\/li>\n<li>Inhibition of return refers to the phenomenon whereby it is more dif\ufb01cult to reassign focal attention to a place that had been attended approximately 600-900ms earlier than assign it to a new location. Recent studies have shown what is inhibited is not primarily the location that had been earlier attended, but the object that was at that location. p188<\/li>\n<li>Whenever some set of properties is perceived as a unitary whole, the resulting hole has many of the properties of a perceptual &#8220;object&#8221;, insofar as it can be tracked through continuous changes (in properties, not in space) and shows single-object superiority effects. It appears, therefore, that &#8220;objecthood&#8221; need not be de\ufb01ned in terms of spatial coherence or spatiotemporal continuity. p193<\/li>\n<\/ul>\n<h2>The link between vision and the world : visual indexes<\/h2>\n<ul>\n<li>What distinguishes an object as an individual, separate from the other features in a scene, is a purely causal sequence of events beginning with some property or combination of properties of the scene and eventuating in an index being grabbed from a pool of potential indexes. Until some piece of the visual \ufb01eld gets segregated and &#8220;picked out&#8221;, or made accessible, in this way, no visual operation can be applied to it, since it does not exist as something distinct from the entire \ufb01eld. Thus, to recognize or otherwise analyze a visual object in the world, we must \ufb01rst distinguish it as a primitive individual thing, separate from the other clutter in the visual \ufb01eld. p210<\/li>\n<li>Just as the separation of \ufb01gure from ground is a primitive function of visual architecture, so also is tracking. p212<\/li>\n<li>Whatever the basis for the solution to the correspondence problem in particular cases, the solution de\ufb01nes what counts as a primitive visual object for the visual system. p212<\/li>\n<li>We have evolved a basic mechanism whose function, within its domain of application (or its &#8220;echological niche&#8221;), is close enough to that of identity maintainance that it can serve as a surrogate for full individuation and identity maintainance. Such a mechanism maintains the correct identity correspondences in most circumstances of importance to us, situations that are typical in our kind of world. Because such a mechanism operates without regard to knowledge and expectation, it is able to operate quickly. p214<\/li>\n<li>In the deictic strategy, people encode only those aspects of the visual scene that are relevant to a particular task, and they rely on a deictic reference mechanism to record more as the need arises. p255<\/li>\n<li>Visual-index theories does not prohibit temporal properties, or any other properties, for in\ufb02uencing what can or cannot be tracked. But since the tracking process is hypothesized to be non-conceptual, how objects are interpreted (what they are seen as) should not affect whether they are tracked. p265<\/li>\n<\/ul>\n<h2>The puzzle of mental imagery<\/h2>\n<ul>\n<li>Even if we do \ufb01nd that there are properties intrinsic to reasoning with images, we must still ask whether these arise because certain imagery-speci\ufb01c forms of representation or processing are being deployed or, as is generally claimed, because imagery involves applying speci\ufb01cally visual mechanisms to perceive an inner picture-like (depictive) pattern. p290<\/li>\n<li>When we speak of a representation, there are at least three levels of analysis at which we can theorize. At the \ufb01rst level we can ask about the representation\u2019s content, or what the representation represents &#8211; what it is about. At the second level of analysis, we can inquire about the form of the representation, the system of codes by which mental states can represent aspects of the world. The third level of analysis is concerned with how representations are realized in biological tissues. p290-292<\/li>\n<li>The task of imagining something unfolding in your mind\u2019s eye is a special task: it\u2019s the task of simulating as many aspects of the visual situation as possible, as many aspects as you can &#8211; not because you are being led on by the experimenter and not because of some special property of mental imagery, but because this is what it means to &#8220;imagne X happening&#8221;. p300<\/li>\n<li>The experience of imaging is not an experience that reveals the form of the image ; rather, it reveals the content of the image. The experience of the image tells us what it&#8217;s an image of. p351<\/li>\n<\/ul>\n<h2>The search for a spatial display in the brain<\/h2>\n<ul>\n<li>By picking out (i.e., indexing) certain objects in the world, and by binding certain thoughts to these objects, you are able to draw conclusions about their con\ufb01guration by visually noticing these properties. p377<\/li>\n<li>There is considerable evidence that the visuomotor system is itself an encapsulated system, which, like the early-vision system, is able to respond only to information arriving from the eyes, often including visual information that is not available to consciousness. As with the visual system, only certain limited kinds of modulations of its characteristic behavior can be imposed by cognition. p411<\/li>\n<li>The fact that mental images represent the conceptual content of a scene (either recalled from memory or constructed during certain tasks) explains why images are distorted over time in characteristic ways, why they can\u2019t be visually (re)interpreted, and why they can fail to be determinate in ways that no picture can fail to be determinate. p412<\/li>\n<li>No visual stimulus carries information about the picture\u2019s interpretation. By contrast, mental images are the interpretation and are therefore conceptual. p414<\/li>\n<li>Image representations contain information about the appearance of things, so they use the vocabulary of visual properties (e.g., color, brightness, shape, texture and so on). p423<\/li>\n<li>Image representations contain information about the relative location of things, so they use the vocabulary of geometrical relations (above, inside, beside, to the right of and so on). p423<\/li>\n<li>Image representations typically refer to individual things ; they represent token individuals (things or objects, or whatever early vision delivers). p424<\/li>\n<li>Image representations lack explicit quanti\ufb01ers, disjunctions and negation. p424<\/li>\n<\/ul>\n<h2>Visual thinking<\/h2>\n<ul>\n<li>The usefulness of diagrams, graphs, charts, and other visual devices relies on the fact that people are good at visually detecting certain geometrical relations. p445<\/li>\n<li>Vision appears to keep track of where information is located in the world rather than encoding it all in memory, thus in effect providing what amounts to a very large memory. p445<\/li>\n<li>Vision routinely goes beyond the information given; a percept is invariably a generalization of the individual properties of a unique stimulus.<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Zenon Pylyshyn The puzzle of seeing Very little information is retained from one glance to another when the eyes move, or even when the eyes do not move but when the display disappears brie\ufb02y. &#8211; p18 Information about a visual scene is not stored in pictorial form, but rather is stored in a form more &#8230; <a title=\"Seeing and Visualizing, it\u2019s not what you think\" class=\"read-more\" href=\"https:\/\/www.labri.fr\/perso\/barla\/blog\/?p=4821\" aria-label=\"Read more about Seeing and Visualizing, it\u2019s not what you think\">Read more<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[621],"tags":[],"class_list":["post-4821","post","type-post","status-publish","format-standard","hentry","category-books"],"_links":{"self":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts\/4821","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4821"}],"version-history":[{"count":5,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts\/4821\/revisions"}],"predecessor-version":[{"id":22811,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=\/wp\/v2\/posts\/4821\/revisions\/22811"}],"wp:attachment":[{"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4821"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4821"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.labri.fr\/perso\/barla\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4821"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}