Portal de Periódicos da CAPES

Brain States: Top-Down Influences in Sensory Processing

2007; Cell Press; Volume: 54; Issue: 5 Linguagem: Inglês

10.1016/j.neuron.2007.05.019

1097-4199

Charles D. Gilbert, Mariano Sigman,

Visual perception and processing mechanisms

All cortical and thalamic levels of sensory processing are subject to powerful top-down influences, the shaping of lower-level processes by more complex information. New findings on the diversity of top-down interactions show that cortical areas function as adaptive processors, being subject to attention, expectation, and perceptual task. Brain states are determined by the interactions between multiple cortical areas and the modulation of intrinsic circuits by feedback connections. In perceptual learning, both the encoding and recall of learned information involves a selection of the appropriate inputs that convey information about the stimulus being discriminated. Disruption of this interaction may lead to behavioral disorders, including schizophrenia. All cortical and thalamic levels of sensory processing are subject to powerful top-down influences, the shaping of lower-level processes by more complex information. New findings on the diversity of top-down interactions show that cortical areas function as adaptive processors, being subject to attention, expectation, and perceptual task. Brain states are determined by the interactions between multiple cortical areas and the modulation of intrinsic circuits by feedback connections. In perceptual learning, both the encoding and recall of learned information involves a selection of the appropriate inputs that convey information about the stimulus being discriminated. Disruption of this interaction may lead to behavioral disorders, including schizophrenia. Though neuroscientists are beginning to establish how the activation of cortical regions and the responses of cortical neurons correlate with behaviors, the enduring mystery is what is the nature of a brain state, the fundamental algorithm, at the level of cortical circuitry, by which cognition arises. To derive this algorithm, one must analyze brain circuits in a behavioral context. The classical view of information processing in the brain is based on a hierarchical organization. In the visual system, pathways start from the analysis of very simple, local attributes, and representation of visual information becomes progressively more complex as one moves up the hierarchy. However, from a computational point of view, it is unlikely that feedforward mechanisms alone can achieve flexible and invariant pattern recognition in a complex and rapidly changing environment. Recent findings have changed radically the view of the role, range, and functional diversity of top-down interactions in the cortex. We have learned that the function of any area of the cerebral cortex, including that of primary visual cortex, is subject to top-down influences of attention, expectation, and perceptual task. Internal representations of the world, acquired by experience, affect our brain's strategy for analyzing visual scenes. Vision is an active process, and the function of any cortical area is not fixed—each area runs different “programs” according to context and to the current perceptual requirements. Visual processing therefore involves countercurrent streams of information flow, and the operation of primary visual cortex involves an interaction between bottom-up information coming from the retina and feedback connections coming from higher-order cortical areas. The general idea of top-down influence is that complex information that is represented at higher stages of processing influences simpler processes occurring at antecedent stages. Whereas some of the earlier work on spatial attention—the most studied instance of top-down modulation—suggested that significant influences of attention are found only at high levels in the visual pathway, it is becoming increasingly clear that even at the earliest stages in cortical sensory processing the functional properties of neurons are subject to influences of attention, as well as other forms of top-down modulation. The view of the perceptual role of attention has gone beyond the simplistic metaphor of attention acting as a “spotlight.” The notion of attention itself may be inadequate as a descriptor of the full range of top-down influences that are exerted. Top-down influences have been shown to operate over a large variety of categories, including features, surfaces, objects, object categories, temporal context, and virtually any other perceptual group. Furthermore, the effect of top-down processing is not best represented as that resulting from a spotlight. Instead, the effects can be of many different kinds, not only augmenting or multiplying responses but also sharpening tuning curves, controlling contextual influences, or acting as a modulator of plasticity. One therefore has to either expand the definition of attention or describe a range of top-down influences that extend beyond the conventional use of the term. The emerging evidence suggests that any cortical area is an adaptive processor. Rather than performing a fixed and stereotyped operation on input coming from the retina, it makes different calculations according to the immediate sensory and behavioral context. This moment-by-moment functional switching is likely mediated by an interaction between feedback connections from higher- to lower-order cortical areas and intrinsic cortical circuits. The role of top-down influences is then to set the cortex in a specific working mode according to behavioral requirements that are updated dynamically. In effect, these ideas reverse the central dogma of sensory processing, with a flow of information from higher- to lower-order cortical areas playing a role equal in importance to the feedforward pathways. The construction of a subjective percept involves making the best sense of sensory inputs based on a set of hypotheses or constraints derived by prior knowledge and contextual influences. Conversely, the top-down expectations and hypotheses are set by feedforward information, the sensory evidence. Under this view, there is no starting point for information flow. Even the prefrontal cortex, arguably the highest-order area in hierarchical views, can be set in different modes depending on task requirements. A strictly linear hierarchy would leave the highest levels without a source of top-down influences, so is it reasonable to speak at all of a starting point of information flow? Rather, in accordance with other theories of brain function (Mackay, 1956Mackay D.M. Towards an information-flow model of human behaviour.Br. J. Psychol. 1956; 47: 30-43Crossref PubMed Google Scholar, Sporns et al., 1991Sporns O. Tononi G. Edelman G.M. Modeling perceptual grouping and figure-ground segregation by means of active reentrant connections.Proc. Natl. Acad. Sci. USA. 1991; 88: 129-133Crossref PubMed Scopus (109) Google Scholar, Mumford, 1992Mumford D. On the computational architecture of the neocortex. II. The role of cortico-cortical loops.Biol. Cybern. 1992; 66: 241-251Crossref PubMed Scopus (375) Google Scholar, Ullman, 1995Ullman S. Sequence seeking and counter streams: a computational model for bidirectional information flow in the visual cortex.Cereb. Cortex. 1995; 5: 1-11Crossref PubMed Google Scholar, Ullman, 1996Ullman S. High-Level Vision. Object Recognition and Visual Cognition. The MIT Press, Cambridge, MA1996Google Scholar, Di Lollo et al., 2000Di Lollo V. Enns J.T. Rensink R.A. Competition for consciousness among visual events: the psychophysics of reentrant visual processes.J. Exp. Psychol. Gen. 2000; 129: 481-507Crossref PubMed Google Scholar, Roelfsema et al., 2000Roelfsema P.R. Lamme V.A. Spekreijse H. The implementation of visual routines.Vision Res. 2000; 40: 1385-1411Crossref PubMed Scopus (77) Google Scholar, Rowe et al., 2002Rowe J. Friston K. Frackowiak R. Passingham R. Attention to action: specific modulation of corticocortical interactions in humans.Neuroimage. 2002; 17: 988-998Crossref PubMed Google Scholar, Dehaene et al., 2003bDehaene S. Sergent C. Changeux J.P. A neuronal network model linking subjective reports and objective physiological data during conscious perception.Proc. Natl. Acad. Sci. USA. 2003; 100: 8520-8525Crossref PubMed Scopus (308) Google Scholar, Lee and Mumford, 2003Lee T.S. Mumford D. Hierarchical Bayesian inference in the visual cortex.J. Opt. Soc. Am. A Opt. Image Sci. Vis. 2003; 20: 1434-1448Crossref PubMed Google Scholar, Deco and Rolls, 2004Deco G. Rolls E.T. A neurodynamical cortical model of visual attention and invariant object recognition.Vision Res. 2004; 44: 621-642Crossref PubMed Scopus (145) Google Scholar, Deco and Rolls, 2005Deco G. Rolls E.T. Attention, short-term memory, and action selection: a unifying theory.Prog. Neurobiol. 2005; 76: 236-256Crossref PubMed Scopus (152) Google Scholar), we propose that perception results from a reverberation (or resonance) between feedforward and top-down information. The ignition of such reverberation may differ in different contexts from an external stimulus to an internal state. In this view, the brain goes through a succession of brain states, with each state serving as the source of top-down influences for the subsequent state. In this dynamic process, task requirements and hypothesis setting are updated by sensory evidence, which in turn causes cortical areas to execute different programs. We propose that this may have an important role in learning and plasticity, postulating that perceptual learning may involve linking the appropriate intrinsic connections to the feedback signal associated with a particular task. Here we will consider the kinds of information that may be conveyed by top-down interactions. The higher-order information may include learned, internal representations of the shapes of objects and of the abstract syntax of object relationships. It may also include information about behavioral context, which would include attention, expectation, and perceptual task. We will also discuss ideas about how disruption of top-down influences, or disconnection of cortical interactions, may play a role in psychiatric disorders such as schizophrenia. A classical view of top-down influences was given by the Gestalt psychologists early in the 20th century. They emphasized that the perception of objects was not achieved by an assembly of the parts of objects but rather that perception was based on holistic patterns. This was expressed most succinctly by Max Wertheimer, who observed that “There are entities where the behavior of the whole cannot be derived from its individual elements nor from the way these elements fit together; rather the opposite is true: the properties of any of the parts are determined by the intrinsic structural laws of the whole” (Wertheimer, 1938bWertheimer M. Uber Gestalttheorie.in: Ellis W.D. A Source Book of Gestalt Psychology. Harcourt, Brace and Co., New York1938: 1-11Crossref Google Scholar). This is illustrated in the principle of “good continuation,” where one tends to link line segments that are collinear and have similar orientation and not those making an abrupt change in direction (Figure 1A). This property is seen also in contour saliency, where contours made of line segments that have a gradual change in orientation tend to pop out from complex backgrounds, in contrast to those with random jitter in the orientation in their composite line segments (Figure 1B). The rule of good continuation makes more tractable the problem of how to link the elements of complex scenes into contours belonging to particular objects and segmenting them from the elements of the background. One sees the rules of perceptual organization reflected in the response properties of neurons in primary visual cortex (V1). This is seen in the dependence of neuronal response upon context and the nature of the receptive field. A single oriented line segment will elicit a brisk response from a neuron when the appropriately oriented line is placed within a small part of visual space, that neuron's receptive field. When that line is embedded within a complex background of randomly oriented and positioned line elements, the neuron's response is substantially inhibited. If one shifts line elements from the background into alignment with the line within the receptive field, the neuron's response becomes greatly facilitated (Kapadia et al., 1995Kapadia M.K. Ito M. Gilbert C.D. Westheimer G. Improvement in visual sensitivity by changes in local context: parallel studies in human observers and in V1 of alert monkeys.Neuron. 1995; 15: 843-856Abstract Full Text PDF PubMed Scopus (566) Google Scholar, Kapadia et al., 1999Kapadia M.K. Westheimer G. Gilbert C.D. Dynamics of spatial summation in primary visual cortex of alert monkeys.Proc. Natl. Acad. Sci. USA. 1999; 96: 12073-12078Crossref PubMed Scopus (209) Google Scholar, Li et al., 2006Li W. Piech V. Gilbert C.D. Contour saliency in primary visual cortex.Neuron. 2006; 50: 951-962Abstract Full Text Full Text PDF PubMed Scopus (100) Google Scholar). In effect, the response of the neuron is as dependent on the global characteristics of the contour extending well outside the core of the neuron's receptive field as it is on the attributes of the line segment within the receptive field. This kind of effect is termed a “contextual” influence. Contextual influences, exerted on neuronal responses, have been implicated in a number of perceptual functions, including contour integration, surface segmentation, color constancy, and object motion. To what extent are these contextual influences “top-down” versus feedforward, and can one equate these influences with feedback connections from higher- to lower-order cortical areas, or are they derived from connections that are intrinsic to the areas where they are found? The predominant view of cortical sensory processing is that as one proceeds along the visual pathway from primary visual cortex at the occipital pole to higher-order visual areas in the temporal lobe, neurons become selective to progressively more complex stimuli—“complexification.” Early areas are thought to analyze simple attributes, such as orientation, direction of movement, or color, and this analysis is limited to local features within a very restricted window. Higher areas assemble the local stimulus features into more complex shapes, and they integrate information over progressively larger parts of visual space. But both anatomical and physiological evidence shows that even at the earliest stages of cortical processing neurons can integrate information over large areas and that they can be endowed with selectivity for complex shapes. The belief that a particular property comes from a single source, however, is likely to be an oversimplification. The expression of one input to a neuron may depend on the state of activation of other inputs. The properties expressed by a given neuron or a given cortical area may be a function of an interaction between different cortical areas (McIntosh, 1999McIntosh A.R. Mapping cognition to the brain through neural interactions.Memory. 1999; 7: 523-548Crossref PubMed Google Scholar, McIntosh, 2000McIntosh A.R. Towards a network theory of cognition.Neural Netw. 2000; 13: 861-870Crossref PubMed Scopus (223) Google Scholar), and more specifically, between local circuits in an area and feedback and feedforward connections from other cortical areas. Top-down influences are sometimes equated with attention, and attention is often thought of in terms of spatial attention, the location of attentional focus. But spatial attention is just one of many forms of attentional influence, and just about anything can be attended, including objects, features (such as orientation or color), motor actions, and time. Other top-down influences include perceptual task, priming, expectation, and hypothesis testing. Parsing the forms of top-down and attentional influence into different categories is not straightforward, since there is an overlap between them and some of the differences may be merely semantic. It is difficult, for example, to separate object expectation from object-oriented attention, or perceptual task from feature-oriented attention. The important thing to note is the amount of information carried by top-down influences as a whole. They do not represent only a spatial coordinate, but the rich diversity of one's internal representations of object identity and task sequencing. Attention is also not an all-or-none phenomenon, but can be graded in intensity. In fact, one of the difficulties in studies on attention is to ensure that an unattended stimulus is in fact unattended, because it is difficult to drive attention to zero unless the task at the attended location is highly demanding of attentional resources (Joseph et al., 1997Joseph J.S. Chun M.M. Nakayama K. Attentional requirements in a ‘preattentive’ feature search task.Nature. 1997; 387: 805-807Crossref PubMed Google Scholar). Another reason why the attentional spotlight is not the best metaphor is that attention has properties that encompass attended objects. In a way, the focus of attention fills the boundaries of the attended object and is therefore referred to as object-oriented attention. Attention can also be directed toward a feature, such as color or orientation, and as such is distributed across the visual field. There are other potential forms of top-down control, however. These include perceptual task, where the discrimination or detection task that is performed at the attended location affects the ways in which the visual stimulus is processed. Expectation may also play a role, whereby internal representations of objects can influence how scenes are segmented. This may represent a form of hypothesis testing, such that before objects are identified, the visual system compares stored representations of object forms against bottom-up information on stimulus characteristics. A dramatic demonstration of the specificity of top-down influences is seen in priming. One can construct an image containing an embedded figure that is nearly impossible to interpret (Figure 2; Porter, 1954Porter P.B. Another picture puzzle.Am. J. Psychol. 1954; 67: 550-551Crossref Google Scholar). If one briefly views a more fully rendered version of the image (Figure 3) and then views the initial, ambiguous version of the same image, the figure becomes immediately apparent. Thus expectation of a particular figure contributes to figure/ground segregation. The role of object expectation is also seen in the classic vase/face ambiguous figure, which can be consciously shifted from one object percept to another (Rubin, 1915Rubin E. Synsoplevede Figurer: Studien I psykologisk analyse, German translation 1921: Visuell wahrgenommene Figuren: Studient in psychologischer Analyse edn. Glydendalske, Copenhagen1915Google Scholar). Computational models of scene segmentation that utilize top-down representation of object shape work much better than segmentation models that rely on bottom-up mechanisms (Ullman, 1995Ullman S. Sequence seeking and counter streams: a computational model for bidirectional information flow in the visual cortex.Cereb. Cortex. 1995; 5: 1-11Crossref PubMed Google Scholar, Deco and Rolls, 2004Deco G. Rolls E.T. A neurodynamical cortical model of visual attention and invariant object recognition.Vision Res. 2004; 44: 621-642Crossref PubMed Scopus (145) Google Scholar, Deco and Rolls, 2005Deco G. Rolls E.T. Attention, short-term memory, and action selection: a unifying theory.Prog. Neurobiol. 2005; 76: 236-256Crossref PubMed Scopus (152) Google Scholar).Figure 3This Image Primes Subjects' Ability to Segment and Recognize the Form Shown in Figure 2View Large Image Figure ViewerDownload Hi-res image Download (PPT) Several lines of evidence support the idea that attention can be directed to an entire object (for reviews see Driver and Baylis, 1998Driver J. Baylis G.C. Attention and visual object segmentation.in: Parasuraman R. The Attentive Brain. MIT Press, Cambridge1998: 299-325Google Scholar, Scholl et al., 2001Scholl B.J. Pylyshyn Z.W. Feldman J. What is a visual object? Evidence from target merging in multiple object tracking.Cognition. 2001; 80: 159-177Crossref PubMed Scopus (146) Google Scholar). Two judgments that concern the same object can be made simultaneously without loss of accuracy, whereas two judgments that concern different objects cannot (Duncan, 1984Duncan J. Selective attention and the organization of visual information.J. Exp. Psychol. Gen. 1984; 113: 501-517Crossref PubMed Scopus (1028) Google Scholar). Moreover, it is easer to divide attention between elements of the same, rather than different, perceptual groups (Baylis and Driver, 1993Baylis G.C. Driver J. Visual attention and objects: evidence for hierarchical coding of location.J. Exp. Psychol. Hum. Percept. Perform. 1993; 19: 451-470Crossref PubMed Scopus (243) Google Scholar, Egly et al., 1994Egly R. Driver J. Rafal R.D. Shifting visual attention between objects and locations: evidence from normal and parietal lesion subjects.J. Exp. Psychol. Gen. 1994; 123: 161-177Crossref PubMed Google Scholar, Watson and Kramer, 1999Watson S.E. Kramer A.F. Object-based visual selective attention and perceptual organization.Percept. Psychophys. 1999; 61: 31-49Crossref PubMed Google Scholar, Lamy and Egeth, 2002Lamy D. Egeth H. Object-based selection: the role of attentional shifts.Percept. Psychophys. 2002; 64: 52-66Crossref PubMed Google Scholar). One demonstration of coherent-motion-defined object-oriented attention is obtained at the surface segmentation level. Two overlapping patterns of dots rotating in opposite directions yield the percept of two superimposed transparent surfaces. Attending to one set of dots impairs subjects' ability to make judgments about the other set. Since the two sets of dots are overlapping, the cueing effect cannot be explained by spatial attention (Valdes-Sosa et al., 1998Valdes-Sosa M. Cobo A. Pinilla T. Transparent motion and object-based attention.Cognition. 1998; 66: B13-B23Crossref PubMed Google Scholar, Valdes-Sosa et al., 2000Valdes-Sosa M. Cobo A. Pinilla T. Attention to object files defined by transparent motion.J. Exp. Psychol. Hum. Percept. Perform. 2000; 26: 488-505Crossref PubMed Google Scholar, Rodriguez et al., 2002Rodriguez V. Valdes-Sosa M. Freiwald W. Dividing attention between form and motion during transparent surface perception.Brain Res. Cogn. Brain Res. 2002; 13: 187-193Crossref PubMed Scopus (22) Google Scholar). This is true even when the color of the two moving surfaces is identical, implying that attention is not directed to space or color, but rather to the whole surface (Mitchell et al., 2003Mitchell J.F. Stoner G.R. Fallah M. Reynolds J.H. Attentional selection of superimposed surfaces cannot be explained by modulation of the gain of color channels.Vision Res. 2003; 43: 1323-1328Crossref PubMed Scopus (23) Google Scholar). When the two surfaces are presented in a rapid succession, following a classical attentional blink paradigm, perception of the first surface severely limits the ability to perceive the second (Pinilla et al., 2001Pinilla T. Cobo A. Torres K. Valdes-Sosa M. Attentional shifts between surfaces: effects on detection and early brain potentials.Vision Res. 2001; 41: 1619-1630Crossref PubMed Scopus (31) Google Scholar). Finally, it has been shown that when the two surfaces are presented in binocular rivalry, directing attention to one of the surfaces enhances its saliency and access to conscious perception (Mitchell et al., 2004Mitchell J.F. Stoner G.R. Reynolds J.H. Object-based attention determines dominance in binocular rivalry.Nature. 2004; 429: 410-413Crossref PubMed Scopus (139) Google Scholar), thus indicating that object-based attention acts as a modulator to conscious selection. Object-level attention modulation of performance has been studied in other forms of bistable perception, showing attentional control over ambiguous figure reversal (Liebert and Burk, 1985Liebert R.M. Burk B. Voluntary control of reversible figures.Percept. Mot. Skills. 1985; 61: 1307-1310Crossref PubMed Google Scholar, Peterson, 1986Peterson M.A. Illusory concomitant motion in ambiguous stereograms: evidence for nonstimulus contributions to perceptual organization.J. Exp. Psychol. Hum. Percept. Perform. 1986; 12: 50-60Crossref PubMed Scopus (20) Google Scholar, Gomez et al., 1995Gomez C. Argandona E.D. Solier R.G. Angulo J.C. Vazquez M. Timing and competition in networks representing ambiguous figures.Brain Cogn. 1995; 29: 103-114Crossref PubMed Scopus (36) Google Scholar, Toppino, 2003Toppino T.C. Reversible-figure perception: mechanisms of intentional control.Percept. Psychophys. 2003; 65: 1285-1295Crossref PubMed Google Scholar). Neuropsychological evidence also shows that perceptual groups define units of attentional selection. For example, in neglect (considered a landmark of spatial attention, since it normally impairs patients from seeing a portion of the visual field), the boundary between the attended and neglected visual field locations can be determined by the midline of an attended object, as opposed to the visual field midline (Driver et al., 1994Driver J. Baylis G.C. Goodrich S.J. Rafal R.D. Axis-based neglect of visual shapes.Neuropsychologia. 1994; 32: 1353-1365Crossref PubMed Scopus (96) Google Scholar, Tipper and Behrmann, 1996Tipper S.P. Behrmann M. Object-centered not scene-based visual neglect.J. Exp. Psychol. Hum. Percept. Perform. 1996; 22: 1261-1278Crossref PubMed Google Scholar, Behrmann and Plaut, 2001Behrmann M. Plaut D.C. The interaction of spatial reference frames and hierarchical object representations: evidence from figure copying in hemispatial neglect.Cogn. Affect. Behav. Neurosci. 2001; 1: 307-329Crossref PubMed Google Scholar). Also, it has been shown that spatial extinction in a parietally damaged patient was less severe when bilateral stimuli formed a common surface, even if this required visual filling-in to yield illusory Kanizsa figures or completion of partially occluded figures (Mattingley et al., 1997Mattingley J.B. Davis G. Driver J. Preattentive filling-in of visual surfaces in parietal extinction.Science. 1997; 275: 671-674Crossref PubMed Scopus (167) Google Scholar). These two examples, however, have a somewhat ambiguous interpretation. It can be argued that spatial selection is the dominant mechanism of top-down control and that the object merely defines the region in space to which attention has to be directed. Additional evidence that spatial attention cannot account for all forms of top-down control comes from experiments in which simultaneous attention to multiple object features (beyond form or spatial extent) are studied. An object can be defined as a cluster of features, grouping elements of different dimensions: color, space, orientation, and so on. In this description, an object evolving in time (moving, deforming, rotating…) can be seen as a trajectory in feature space. An important demonstration of object-directed attention involves experiments in which subjects are asked to track an evolving object, changing in three different feature dimensions: color, space, and spatial frequency. Attending to a feature of an object enhances one's ability to discriminate its other features (Blaser et al., 2000Blaser E. Pylyshyn Z.W. Holcombe A.O. Tracking an object through feature space.Nature. 2000; 408: 196-199Crossref PubMed Scopus (141) Google Scholar), thus showing that performance enhancement transfers to judgments of multiple dimensions of the same object. An important effort has been devoted to understanding what subsets of feature space may be grouped as an object and how this relates to perceptual groups that may be attended. The Gestalt psychologists established a series of fundamental principles that govern perceptual grouping (Wertheimer, 1938aWertheimer M. Laws of Organization in Perceptual Forms. Harcourt, Brace & Jovanovitch, London1938Crossref Google Scholar). However, gestaltian principles do not fully account for the elements in feature space to which attention can be guided (Scholl et al., 2001Scholl B.J. Pylyshyn Z.W. Feldman J. What is a visual object? Evidence from target merging in multiple object tracking.Cognition. 2001; 80: 159-177Crossref PubMed Scopus (146) Google Scholar). Grouping operations depend on the identification of objects and object parts, which may vary on many geometric and semantic factors (i.e., a door may be a whole object made of pieces, or a part of a house made of door, windows, etc…). The versatile and intrinsically dynamic nature of grouping operations is demonstrated by the Marroquin pattern (Marroquin, 1976Marroquin J.L. Human Visual Perception of Structure. MIT Press, Cambridge, MA1976Google Scholar, Marr, 1982Marr D. Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. Freeman, San Francisco1982Google Scholar) illustrated in Figure 4. When attending to an invariant static pattern, circular shapes appear and vanish dynamically at various locations. Following Kofka (Kofka, 1935Kofka K. Principles of Gestalt Psychology. Harcourt,Brace, New York1935Google Scholar), Palmer and Rock, 1994Palmer S.E. Rock L. Rethinking perceptual organization: The role of uniform connectedness.Psychol. Bull. 1994; 1: 29-55Crossref Scopus (296) Google Scholar proposed a scheme for perceptual organization that incorporates the Gestaltian principles of grouping, based on hierarchies of objects and object parts. A first categorization of object into parts is based on regions of uniformly connected (UC) visual properties, such as luminance, color, or texture. These regions, in turn, can be parsed into branches whose